refactor(vo): address final review — accurate docstring + update_depth_hint tests

Final review findings (Important):
- I1: e2e test docstring overclaimed — harness always uses ORBVisualOdometry.
  Rewrite docstring to describe the actual scope: smoke test + ORB regression
  guard. Wiring Mono-Depth wrapper through the harness is a sprint 2 task.
- I2: update_depth_hint had no tests. Add 2 tests: clamp at 1.0m for bogus
  values, and verify next compute_relative_pose uses the updated scale.
- I3: add TODO marker for sprint 2 deduplication with CuVSLAMVisualOdometry.

No behavior change — only docstrings, TODO markers, and test coverage.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

tests/test_vo.py

@@ -301,3 +301,43 @@ def test_mono_depth_create_vo_backend_selects_it():
     from gps_denied.core.vo import CuVSLAMMonoDepthVisualOdometry, create_vo_backend
     vo = create_vo_backend(prefer_mono_depth=True, depth_hint_m=600.0)
     assert isinstance(vo, CuVSLAMMonoDepthVisualOdometry)
+
+
+def test_mono_depth_update_depth_hint_clamps_below_one():
+    """update_depth_hint clamps bogus/negative barometer values to minimum 1.0m."""
+    from gps_denied.core.vo import CuVSLAMMonoDepthVisualOdometry
+    vo = CuVSLAMMonoDepthVisualOdometry(depth_hint_m=600.0)
+    vo.update_depth_hint(-5.0)
+    assert vo._depth_hint_m == 1.0
+    vo.update_depth_hint(0.0)
+    assert vo._depth_hint_m == 1.0
+    vo.update_depth_hint(0.5)
+    assert vo._depth_hint_m == 1.0
+
+
+def test_mono_depth_update_depth_hint_affects_subsequent_scale():
+    """update_depth_hint changes scale used by next compute_relative_pose call."""
+    from unittest.mock import patch
+    from gps_denied.core.vo import CuVSLAMMonoDepthVisualOdometry, ORBVisualOdometry
+    from gps_denied.schemas.vo import RelativePose
+
+    cam = CameraParameters(
+        focal_length=16.0, sensor_width=23.2, sensor_height=17.4,
+        resolution_width=640, resolution_height=480,
+    )
+    img = np.zeros((480, 640), dtype=np.uint8)
+    unit_pose = RelativePose(
+        translation=np.array([1.0, 0.0, 0.0]),
+        rotation=np.eye(3),
+        confidence=1.0, inlier_count=100, total_matches=100,
+        tracking_good=True, scale_ambiguous=True,
+    )
+
+    with patch.object(ORBVisualOdometry, "compute_relative_pose", return_value=unit_pose):
+        vo = CuVSLAMMonoDepthVisualOdometry(depth_hint_m=600.0)
+        pose_before = vo.compute_relative_pose(img, img, cam)
+        vo.update_depth_hint(1200.0)
+        pose_after = vo.compute_relative_pose(img, img, cam)
+
+    assert np.allclose(pose_before.translation, [1.0, 0.0, 0.0])  # 600/600 = 1.0x
+    assert np.allclose(pose_after.translation, [2.0, 0.0, 0.0])   # 1200/600 = 2.0x