feat!: reorganize detection and tracking pipeline

Refactor the package into common, schema, detection, and tracking namespaces and move dataset-specific ActualTest utilities into tests/support.

Add a pluggable detection stack with typed protocols, pydantic-settings config, loguru-based runner logging, cvmmap and headless video sources, NATS and parquet sinks, and a structured coco-wholebody133 payload path.

Teach tracking replay loading to consume parquet detection directories directly, preserve empty frames, and keep the video-to-parquet-to-tracking workflow usable for offline E2E runs.

Vendor the local mmcv and xtcocotools wheels under Git LFS, update uv sources/lock state, and refresh the mmcv build so mmcv.ops loads successfully with the current torch+cu130 environment.

tests/test_camera_conventions.py

@@ -8,9 +8,9 @@ import pytest
 
 pytest.importorskip("rpt")
 
-from pose_tracking_exp.models import CameraCalibration, SceneConfig
-from pose_tracking_exp.replay import load_scene_file
-from pose_tracking_exp.rpt_adapter import build_rpt_config
+from pose_tracking_exp.schema import CameraCalibration, CameraModel, SceneConfig, parse_camera_model
+from pose_tracking_exp.tracking.replay_io import load_scene_file
+from pose_tracking_exp.tracking.rpt_adapter import build_rpt_config
 
 
 class _CameraArgs(NamedTuple):
@@ -19,7 +19,7 @@ class _CameraArgs(NamedTuple):
     height: int
     K: np.ndarray
     DC: np.ndarray
-    model: str
+    model: CameraModel
 
 
 def _camera_args() -> _CameraArgs:
@@ -29,7 +29,7 @@ def _camera_args() -> _CameraArgs:
         height=480,
         K=np.asarray([[500.0, 0.0, 320.0], [0.0, 500.0, 240.0], [0.0, 0.0, 1.0]], dtype=np.float64),
         DC=np.zeros(5, dtype=np.float64),
-        model="pinhole",
+        model=parse_camera_model("pinhole"),
     )
 
 
@@ -139,7 +139,7 @@ def test_build_rpt_config_uses_pose_convention(monkeypatch: pytest.MonkeyPatch)
         captured["min_group_size"] = min_group_size
         return captured
 
-    monkeypatch.setattr("pose_tracking_exp.rpt_adapter.rpt.make_triangulation_config", fake_make_triangulation_config)
+    monkeypatch.setattr("pose_tracking_exp.tracking.rpt_adapter.rpt.make_triangulation_config", fake_make_triangulation_config)
 
     build_rpt_config(scene, min_match_score=0.5, min_group_size=2)