[WIP] Exploring B.Y.O.C.V. Detect ArUco markers from extracted frames, persist ArUco detections, and merge them with existing portal observations

[arshak93]

Detect ArUco markers from extracted frames, persist ArUco detections, and merge them with existing portal observations so loop-closure and manifest outputs can use both marker sources without breaking legacy scans.

Summary

• add utils/aruco_detection.py to detect ArUco markers from extracted frames, estimate 6-DoF marker poses with camera intrinsics, convert to OpenGL coordinates, and write ArUcoDetections.csv
• invoke ArUco detection in local refinement after process_frames() and before load_scan() so detections are available to loop-closure inputs
• merge server-side ArUco detections additively in load_scan() while preserving DMT priority (PortalDetections.csv then Observations.csv) and improving resilience when legacy portal files are missing
• fill portal_sizes for aruco_* IDs using the default marker size so downstream portal CSV and manifest generation remain valid