Implement Phase 1: full data pipeline, model, training, and config support

- Data pipeline: loader (parquet→numpy), transforms (log, local-frame
  Rodrigues rotation, Normalizer), StepsDataset with event-ID-based split
- Model: SinusoidalEmbedding, ConditionEncoder, ResBlock, DenoisingMLP
- Schedule: cosine DDPM and conditional flow matching loss (Lipman 2022)
- Samplers: flow (Euler ODE), DDPM ancestral, DDIM deterministic
- Training loop: AdamW + cosine LR, grad clipping, best-val checkpoint
- Validation: per-dimension marginal summary (normalised space)
- CLI: TOML config support with CLI-overrides; hyperparam-encoded output
  directory; config.toml with git hash saved into each run's checkpoint dir
- 21 unit tests covering transforms, network, flow/DDPM losses, dataset splits

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

tests/test_transforms.py

@@ -0,0 +1,66 @@
+import numpy as np
+import pytest
+from giant.data.transforms import (
+    inv_log_transform,
+    local_frame_rotation,
+    log_transform,
+    Normalizer,
+)
+
+
+def test_log_transform_invertible():
+    x = np.array([0.1, 1.0, 10.0, 1000.0], dtype=np.float32)
+    np.testing.assert_allclose(inv_log_transform(log_transform(x)), x, rtol=1e-5)
+
+
+def test_local_frame_rotation_noop_when_aligned():
+    N = 8
+    pre_dir = np.tile([0.0, 0.0, 1.0], (N, 1)).astype(np.float32)
+    rng = np.random.default_rng(0)
+    post_dir = rng.standard_normal((N, 3)).astype(np.float32)
+    post_dir /= np.linalg.norm(post_dir, axis=1, keepdims=True)
+    result = local_frame_rotation(pre_dir, post_dir)
+    np.testing.assert_allclose(result, post_dir, atol=1e-5)
+
+
+def test_local_frame_rotation_preserves_angle():
+    """Angle between pre_dir and post_dir must equal angle between ẑ and rotated."""
+    rng = np.random.default_rng(1)
+    N = 200
+    pre_dir = rng.standard_normal((N, 3)).astype(np.float32)
+    pre_dir /= np.linalg.norm(pre_dir, axis=1, keepdims=True)
+    post_dir = rng.standard_normal((N, 3)).astype(np.float32)
+    post_dir /= np.linalg.norm(post_dir, axis=1, keepdims=True)
+
+    rotated = local_frame_rotation(pre_dir, post_dir)
+
+    cos_before = (pre_dir * post_dir).sum(axis=1)
+    cos_after = rotated[:, 2]  # dot with ẑ = z-component (unit vectors)
+    np.testing.assert_allclose(cos_after, cos_before, atol=1e-5)
+
+
+def test_local_frame_rotation_preserves_norm():
+    rng = np.random.default_rng(2)
+    N = 100
+    pre_dir = rng.standard_normal((N, 3)).astype(np.float32)
+    pre_dir /= np.linalg.norm(pre_dir, axis=1, keepdims=True)
+    post_dir = rng.standard_normal((N, 3)).astype(np.float32)
+    post_dir /= np.linalg.norm(post_dir, axis=1, keepdims=True)
+    result = local_frame_rotation(pre_dir, post_dir)
+    np.testing.assert_allclose(np.linalg.norm(result, axis=1), 1.0, atol=1e-5)
+
+
+def test_normalizer_roundtrip():
+    rng = np.random.default_rng(3)
+    X = rng.standard_normal((200, 9)).astype(np.float32)
+    norm = Normalizer().fit(X)
+    np.testing.assert_allclose(norm.inverse_transform(norm.transform(X)), X, atol=1e-5)
+
+
+def test_normalizer_serialization():
+    rng = np.random.default_rng(4)
+    X = rng.standard_normal((50, 6)).astype(np.float32)
+    norm = Normalizer().fit(X)
+    norm2 = Normalizer.from_dict(norm.to_dict())
+    np.testing.assert_allclose(norm2.mean, norm.mean)
+    np.testing.assert_allclose(norm2.std, norm.std)