Add giant.analysis module for notebook-based model quality diagnostics

Provides stratified marginal comparisons, joint-structure checks (correlation
matrices, physically-coupled pairwise plots, direction alignment), and
physical-constraint validation (unit-norm directions, non-negative raw
targets) for a trained model's generated samples, building on the aggregate
marginal/KL check already in giant.validate.

Supports two entry points: live sampling against a checkpoint + val data
(load_model_bundle/collect_samples), or loading a precomputed
`giant predict --coord local` parquet directly (load_predicted_local) without
needing the checkpoint at all. Predict output is now tagged with parquet
schema metadata so the loader can verify a file's format and reject
coord=global or untagged files with a clear error instead of guessing from
column names.

Also extends the config git-hash mismatch warning (added for --config
loading) to checkpoint loading: both `giant predict` and
analysis.load_model_bundle now look for a config.toml next to the checkpoint
and warn (without failing) if it was generated from a different git commit.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

tests/test_config.py

@@ -0,0 +1,102 @@
+from giant import config as gconfig
+
+
+def _write_config(path, git_hash):
+    path.write_text(
+        f"""
+[train]
+epochs = 5
+
+[model]
+hidden_dim = 64
+
+[meta]
+git_hash = "{git_hash}"
+"""
+    )
+
+
+def test_merge_cli_overrides_applies_file_then_cli(tmp_path, monkeypatch):
+    monkeypatch.setattr(gconfig, "git_hash", lambda: "abc123")
+    path = tmp_path / "config.toml"
+    _write_config(path, "abc123")
+
+    cfg = gconfig.merge_cli_overrides(
+        gconfig.DEFAULT_CONFIG, path, train_overrides={}, model_overrides={"hidden_dim": 128},
+    )
+    assert cfg["train"]["epochs"] == 5  # from file
+    assert cfg["model"]["hidden_dim"] == 128  # CLI override wins over file
+
+
+def test_merge_cli_overrides_warns_on_git_hash_mismatch(tmp_path, monkeypatch, capsys):
+    monkeypatch.setattr(gconfig, "git_hash", lambda: "current999")
+    path = tmp_path / "config.toml"
+    _write_config(path, "old111")
+
+    cfg = gconfig.merge_cli_overrides(gconfig.DEFAULT_CONFIG, path, {}, {})
+
+    assert cfg["train"]["epochs"] == 5  # does not fail, config still applied
+    captured = capsys.readouterr()
+    assert "warning" in captured.err
+    assert "old111" in captured.err
+    assert "current999" in captured.err
+
+
+def test_merge_cli_overrides_no_warning_on_matching_git_hash(tmp_path, monkeypatch, capsys):
+    monkeypatch.setattr(gconfig, "git_hash", lambda: "same123")
+    path = tmp_path / "config.toml"
+    _write_config(path, "same123")
+
+    gconfig.merge_cli_overrides(gconfig.DEFAULT_CONFIG, path, {}, {})
+    assert capsys.readouterr().err == ""
+
+
+def test_merge_cli_overrides_no_warning_when_git_hash_unknown(tmp_path, monkeypatch, capsys):
+    monkeypatch.setattr(gconfig, "git_hash", lambda: "unknown")
+    path = tmp_path / "config.toml"
+    _write_config(path, "abc123")
+
+    gconfig.merge_cli_overrides(gconfig.DEFAULT_CONFIG, path, {}, {})
+    assert capsys.readouterr().err == ""
+
+
+def test_merge_cli_overrides_no_warning_when_meta_section_absent(tmp_path, monkeypatch, capsys):
+    monkeypatch.setattr(gconfig, "git_hash", lambda: "current999")
+    path = tmp_path / "config.toml"
+    path.write_text("[train]\nepochs = 5\n")
+
+    gconfig.merge_cli_overrides(gconfig.DEFAULT_CONFIG, path, {}, {})
+    assert capsys.readouterr().err == ""
+
+
+def test_warn_if_checkpoint_config_mismatch_finds_sibling_toml(tmp_path, monkeypatch, capsys):
+    monkeypatch.setattr(gconfig, "git_hash", lambda: "current999")
+    ckpt_path = tmp_path / "best.pt"
+    ckpt_path.write_bytes(b"")  # contents irrelevant, only its directory is used
+    _write_config(tmp_path / "config.toml", "old111")
+
+    gconfig.warn_if_checkpoint_config_mismatch(ckpt_path)
+
+    captured = capsys.readouterr()
+    assert "warning" in captured.err
+    assert "old111" in captured.err
+    assert "current999" in captured.err
+
+
+def test_warn_if_checkpoint_config_mismatch_no_warning_when_toml_absent(tmp_path, monkeypatch, capsys):
+    monkeypatch.setattr(gconfig, "git_hash", lambda: "current999")
+    ckpt_path = tmp_path / "best.pt"
+    ckpt_path.write_bytes(b"")
+
+    gconfig.warn_if_checkpoint_config_mismatch(ckpt_path)
+    assert capsys.readouterr().err == ""
+
+
+def test_warn_if_checkpoint_config_mismatch_no_warning_when_hashes_match(tmp_path, monkeypatch, capsys):
+    monkeypatch.setattr(gconfig, "git_hash", lambda: "same123")
+    ckpt_path = tmp_path / "best.pt"
+    ckpt_path.write_bytes(b"")
+    _write_config(tmp_path / "config.toml", "same123")
+
+    gconfig.warn_if_checkpoint_config_mismatch(ckpt_path)
+    assert capsys.readouterr().err == ""