OVRTX LiDAR runtime capture

Use ovrtx-sensor-capture to establish multi-sensor creation, selection, stepping, iteration, and cleanup. Then use extend-ovrtx-lidar-capture for the implementation work.

Required technical surface

1. Add a documented sensor/product schema and a typed native builder/decoder for raw returns. Probe capabilities before calling it.
2. Preserve sensor path, RenderVar path, simulation ID, sample time, return ID, coordinate frame, units, and valid/invalid status through every layer.
3. Decode finite positions plus any supported range, intensity, ring/channel, timestamp, or return index without inventing absent fields.
4. Validate an isolated target with measured pose and bounds before a complex scene. Check axis/unit conversion and first-surface behavior numerically.
5. Test no-return, timeout, partial/iterator, terminal error, reset, and cleanup paths. Run one real native capture after unit tests.