Add multi-rollout support to giant analyze (gitea #77)

giant analyze compares N rollout YAMLs against one shared reference file
(all must name the same dataset, checked up front) instead of exactly one
rollout vs one reference, rendering each rollout as its own colored series
against a single reference line/panel. Series names come from a repeated
--label flag, else the YAML stem, else "rollout" for a single YAML — a
single-rollout run keeps rendering identically to before this change.

Bundle now holds a name-keyed dict of rollout sides instead of one fixed
pair, every catalog compute_partial/finalize builds a Reduced.payload
keyed the same way ("series": {name: ...}, "reference": ... as the one
distinguished non-rollout entry), and every renderer draws N series (or
N panels, for the two heatmap-shaped specs and the router/type-embedding
diagnostics, which are inherently one-matrix/one-checkpoint per rollout)
against the reference's fixed dashed-ink style.

tests/test_render.py

@@ -25,41 +25,96 @@ def test_render_router_diagnostics_and_edge_cases(tmp_path: Path):
     reduced = [
         Reduced(
             "rg",
-            "router",
+            "model",
             "router_gating",
             "Router gating",
             "pre-step energy [MeV]",
             {
-                "n_experts": 2,
                 "log_x": True,
-                "router_type": "energy",
-                "rollout": {
-                    "centers": [1.0, 10.0, 100.0],
-                    "means": [[0.6, 0.4], [0.5, 0.5], [0.4, 0.6]],
-                },
-                "reference": {
-                    "centers": [1.0, 10.0, 100.0],
-                    "means": [[0.55, 0.45], [0.5, 0.5], [0.45, 0.55]],
+                "series": {
+                    "flow": {
+                        "n_experts": 2,
+                        "router_type": "energy",
+                        "rollout": {
+                            "centers": [1.0, 10.0, 100.0],
+                            "means": [[0.6, 0.4], [0.5, 0.5], [0.4, 0.6]],
+                        },
+                        "reference": {
+                            "centers": [1.0, 10.0, 100.0],
+                            "means": [[0.55, 0.45], [0.5, 0.5], [0.45, 0.55]],
+                        },
+                    },
+                    "wgan": {
+                        "n_experts": 2,
+                        "router_type": "energy",
+                        "rollout": {"centers": [1.0], "means": [[0.5, 0.5]]},
+                        "reference": {"centers": [1.0], "means": [[0.5, 0.5]]},
+                    },
                 },
             },
         ),
         Reduced(
             "rs",
-            "router",
+            "model",
             "router_share",
             "Router share",
             "species",
             {
-                "categories": ["e-", "gamma"],
-                "n_experts": 2,
-                "router_type": "energy",
-                "rollout": {"e-": [0.7, 0.3], "gamma": [0.2, 0.8]},
-                "reference": {"e-": [0.6, 0.4], "gamma": [0.3, 0.7]},
+                "series": {
+                    "flow": {
+                        "categories": ["e-", "gamma"],
+                        "n_experts": 2,
+                        "router_type": "energy",
+                        "rollout": {"e-": [0.7, 0.3], "gamma": [0.2, 0.8]},
+                        "reference": {"e-": [0.6, 0.4], "gamma": [0.3, 0.7]},
+                    },
+                },
+            },
+        ),
+        Reduced(
+            "rp",
+            "model",
+            "router_share",
+            "Router share by process (reference-only)",
+            "process",
+            {
+                "series": {
+                    "flow": {
+                        "categories": ["compt", "phot"],
+                        "n_experts": 2,
+                        "router_type": "energy",
+                        "reference": {"compt": [0.4, 0.6], "phot": [0.9, 0.1]},
+                    },
+                },
+            },
+        ),
+        Reduced(
+            "rz",
+            "model",
+            "router_specialization",
+            "Router specialization",
+            "pre-step energy [MeV]",
+            {
+                "log_x": True,
+                "series": {
+                    "flow": {
+                        "n_experts": 2,
+                        "chance_level": 0.5,
+                        "rollout": {"centers": [1.0, 10.0], "score": [0.6, 0.7]},
+                        "reference": {"centers": [1.0, 10.0], "score": [0.55, 0.65]},
+                    },
+                    "wgan": {
+                        "n_experts": 4,
+                        "chance_level": 0.25,
+                        "rollout": {"centers": [1.0, 10.0], "score": [0.3, 0.4]},
+                        "reference": {"centers": [], "score": []},
+                    },
+                },
             },
         ),
         Reduced(
             "ru",
-            "router",
+            "model",
             "unavailable",
             "Router unavailable",
             "x",
@@ -73,7 +128,10 @@ def test_render_router_diagnostics_and_edge_cases(tmp_path: Path):
             "x",
             {
                 "edges": [0, 1, 2],
-                "groups": {lbl: {"rollout": [1, 2], "reference": [2, 1]} for lbl in ("a", "b", "c", "d")},
+                "groups": {
+                    lbl: {"series": {"flow": [1, 2], "wgan": [2, 1]}, "reference": [2, 1]}
+                    for lbl in ("a", "b", "c", "d")
+                },
                 "log_y": True,
             },
         ),
@@ -83,7 +141,22 @@ def test_render_router_diagnostics_and_edge_cases(tmp_path: Path):
             "single_hist",
             "Single (log-x)",
             "x",
-            {"edges": [1, 10, 100], "rollout": [5, 1], "log_x": True, "log_y": True},
+            {"edges": [1, 10, 100], "series": {"flow": [5, 1], "wgan": [3, 2]}, "log_x": True, "log_y": True},
+        ),
+        Reduced(
+            "hm",
+            "quality",
+            "heatmap",
+            "Distance summary (2 rollouts)",
+            "grouping axis",
+            {
+                "series": {"flow": [[0.1, 0.2], [0.3, 0.4]], "wgan": [[0.5, 0.6], [0.7, 0.8]]},
+                "row_labels": ["step_length", "edep"],
+                "col_labels": ["overall", "energy"],
+                "cbar_label": "KS statistic",
+                "vmin": 0.0,
+                "vmax": 1.0,
+            },
         ),
     ]
     try:
@@ -117,7 +190,7 @@ def test_render_all_run_gallery_invokes_subprocess(tmp_path: Path, monkeypatch):
             "single_hist",
             "Single",
             "x",
-            {"edges": [0, 1, 2], "rollout": [5, 1]},
+            {"edges": [0, 1, 2], "series": {"rollout": [5, 1]}},
         )
     ]
     for r in reduced:
@@ -142,15 +215,14 @@ def test_render_run_glues_condor_run_meta_into_render_all(tmp_path: Path, monkey
     merge_calls = []
     monkeypatch.setattr(condor_mod, "merge_all", lambda rd: merge_calls.append(Path(rd)))
     meta = condor_mod.RunMeta(
-        rollout="rollout.parquet",
+        rollouts=[{"name": "rollout", "path": "rollout.parquet", "plot_meta": {"checkpoint": "ckpt/best.pt"}}],
         reference="reference.parquet",
         run_dir=str(run_dir),
         title="my-run",
-        plot_meta={"checkpoint": "ckpt/best.pt"},
     )
     monkeypatch.setattr(condor_mod.RunMeta, "load", classmethod(lambda cls, p: meta))
 
-    Reduced("s", "species", "single_hist", "Single", "x", {"edges": [0, 1], "rollout": [1]}).save(
+    Reduced("s", "species", "single_hist", "Single", "x", {"edges": [0, 1], "series": {"rollout": [1]}}).save(
         run_dir / "reduced" / "s.json"
     )
 
@@ -177,7 +249,7 @@ def test_render_one_of_each_kind(tmp_path: Path):
             "x",
             {
                 "edges": [0, 1, 2, 3],
-                "rollout": [1, 2, 3],
+                "series": {"flow": [1, 2, 3], "wgan": [2, 2, 2]},
                 "reference": [3, 2, 1],
                 "log_y": False,
             },
@@ -190,7 +262,7 @@ def test_render_one_of_each_kind(tmp_path: Path):
             "x",
             {
                 "edges": [0, 1, 2],
-                "groups": {"a": {"rollout": [1, 2], "reference": [2, 1]}},
+                "groups": {"a": {"series": {"flow": [1, 2]}, "reference": [2, 1]}},
                 "log_y": False,
             },
         ),
@@ -202,10 +274,8 @@ def test_render_one_of_each_kind(tmp_path: Path):
             "depth",
             {
                 "edges": [0, 1, 2],
-                "rollout_mean": [1, 2],
-                "rollout_std": [0.1, 0.2],
-                "reference_mean": [1.1, 1.9],
-                "reference_std": [0.1, 0.1],
+                "series": {"flow": {"mean": [1, 2], "std": [0.1, 0.2]}},
+                "reference": {"mean": [1.1, 1.9], "std": [0.1, 0.1]},
                 "ylabel": "e",
             },
         ),
@@ -217,7 +287,7 @@ def test_render_one_of_each_kind(tmp_path: Path):
             "species",
             {
                 "labels": ["e-", "gamma"],
-                "rollout": [0.6, 0.4],
+                "series": {"flow": [0.6, 0.4], "wgan": [0.55, 0.45]},
                 "reference": [0.5, 0.5],
                 "ylabel": "frac",
             },
@@ -228,7 +298,20 @@ def test_render_one_of_each_kind(tmp_path: Path):
             "single_hist",
             "Single",
             "x",
-            {"edges": [0, 1, 2], "rollout": [5, 1], "log_y": True},
+            {"edges": [0, 1, 2], "series": {"flow": [5, 1]}, "log_y": True},
+        ),
+        Reduced(
+            "hm1",
+            "secondaries",
+            "heatmap",
+            "Confusion (single rollout)",
+            "predicted",
+            {
+                "series": {"flow": [[1, 0], [0, 1]]},
+                "row_labels": ["0", "1+"],
+                "col_labels": ["0", "1+"],
+                "cbar_label": "count",
+            },
         ),
     ]
     try:
@@ -276,8 +359,8 @@ def test_figure_params_v2_basics_and_router_and_epoch():
         },
         "conditioning": {"particle": {"type": "physical"}},
     }
-    run_meta = {"training_epoch": 12, "best_val_loss": 0.123456, "steps": 10}
-    params = render_mod._figure_params(run_meta | {"model_config": mc})
+    meta = {"training_epoch": 12, "best_val_loss": 0.123456, "steps": 10, "model_config": mc}
+    params = render_mod._figure_params({"rollouts": {"rollout": meta}})
     assert params == {
         "hidden_dim": 256,
         "n_res_blocks": 4,
@@ -297,8 +380,8 @@ def test_figure_params_v2_wgan_reports_noise_dim_not_steps():
             "wgan": {"noise_dim": 32},
         },
     }
-    run_meta = {"model_config": mc, "steps": 10}
-    params = render_mod._figure_params(run_meta)
+    meta = {"model_config": mc, "steps": 10}
+    params = render_mod._figure_params({"rollouts": {"rollout": meta}})
     assert params["mode"] == "wgan"
     assert params["noise_dim"] == 32
     assert "steps" not in params
@@ -309,18 +392,18 @@ def test_figure_params_v2_reports_mode_s2_only_when_it_differs():
         "stage1_model": {"generator": "flow"},
         "stage2_model": {"generator": "flow"},
     }
-    assert "mode_s2" not in render_mod._figure_params({"model_config": same})
+    assert "mode_s2" not in render_mod._figure_params({"rollouts": {"rollout": {"model_config": same}}})
 
     mixed = {
         "stage1_model": {"generator": "flow"},
         "stage2_model": {"generator": "wgan"},
     }
-    params = render_mod._figure_params({"model_config": mixed})
+    params = render_mod._figure_params({"rollouts": {"rollout": {"model_config": mixed}}})
     assert params["mode_s2"] == "wgan"
 
 
 def test_figure_params_old_shape_basics():
-    run_meta = {
+    meta = {
         "model_config": {
             "hidden_dim": 128,
             "n_blocks": 3,
@@ -332,7 +415,7 @@ def test_figure_params_old_shape_basics():
         "best_val_loss": 0.5,
         "steps": 20,
     }
-    params = render_mod._figure_params(run_meta)
+    params = render_mod._figure_params({"rollouts": {"rollout": meta}})
     assert params == {
         "hidden_dim": 128,
         "n_blocks": 3,
@@ -346,20 +429,30 @@ def test_figure_params_old_shape_basics():
 
 
 def test_figure_params_old_shape_wgan_reports_noise_dim_not_steps():
-    run_meta = {
+    meta = {
         "model_config": {"mode": "wgan", "noise_dim": 16},
         "steps": 20,
     }
-    params = render_mod._figure_params(run_meta)
+    params = render_mod._figure_params({"rollouts": {"rollout": meta}})
     assert params["noise_dim"] == 16
     assert "steps" not in params
 
 
+def test_figure_params_multi_rollout_names_the_series():
+    run_meta = {"rollouts": {"flow": {"model_config": {"mode": "flow"}}, "wgan": {"model_config": {"mode": "wgan"}}}}
+    assert render_mod._figure_params(run_meta) == {"rollouts": "flow, wgan"}
+
+
+def test_figure_params_empty_rollouts_is_empty():
+    assert render_mod._figure_params({}) == {}
+    assert render_mod._figure_params({"rollouts": {}}) == {}
+
+
 def test_plot_metadata_includes_note_and_run_meta_parameters():
     r = Reduced("u", "router", "unavailable", "Unavailable", "x", {"note": "no router data"})
-    meta = render_mod._plot_metadata(r, {"title": "run-1", "checkpoint": "ckpt.pt"})
+    meta = render_mod._plot_metadata(r, {"title": "run-1", "reference": "ref.parquet", "rollouts": {"rollout": {}}})
     assert meta["note"] == "no router data"
-    assert meta["parameters"] == {"checkpoint": "ckpt.pt"}
+    assert meta["parameters"] == {"reference": "ref.parquet", "rollouts": {"rollout": {}}}
     assert "title" not in meta["parameters"]