fix: catch cv2.error in points2F(), not just F/mask=None

[petercorke]

Summary

• `points2F()` already handles `cv2.findFundamentalMat()` returning `(None, None)` on degenerate point configurations, converting it to a descriptive `ValueError`. But OpenCV doesn't always fail that cleanly -- on some inputs it raises a raw C++ assertion instead (`Assertion failed: 0 <= _rowRange.start && ... in function 'Mat'`), which isn't caught anywhere.
• Catches `cv2.error` alongside the existing `None` check and re-raises as the same `ValueError`, since it's the same underlying cause (degenerate points), just surfaced differently by OpenCV.
• Found via RVC3-python's `visodom.py` example, a temporal visual-odometry loop that already wraps `estimate(cam.points2F, ...)` in `try/except ValueError` specifically to skip degenerate frames -- the raw `cv2.error` was slipping past that except clause and crashing the whole run at frame 149 of 251.

Test plan

• `pytest tests/ -k "points2F or Camera or camera"` -- 27 passed, 1 skipped
• Ran RVC3-python's `visodom.py` end to end against the bridge dataset (251 frames) -- previously crashed hard at frame 149, now cleanly skips it and completes all 251 frames

[codacy-production]

Up to standards ✅

🟢 Issues 0 issues

Results:
0 new issues

View in Codacy

🟢 Metrics 0 complexity · 0 duplication

Metric	Results
Complexity	0
Duplication	0

View in Codacy

_{NEW Get contextual insights on your PRs based on Codacy's metrics, along with PR and Jira context, without leaving GitHub. Enable AI reviewer
TIP This summary will be updated as you push new changes.}