Remove gold_answer_correct (superseded by answer_correctness 1-4)

Add gold_answer_correct Y/N for dashboard completeness check (derived from answer_correctness>=3)

Collapse 5-point scales to 4-point (5 -> 4 across clarity/correctness/timestamp)

Fix 2 remaining mp=N entries; remap timestamp_correctness 1->2, 2->3

Update annotations from annotator

Update annotations from annotator

Phase 3: merged re-eval scores for 198 flagged pairs; v7 QA fixes applied

Update annotations from annotator (500 videos × 4 splits = 2000 entries)

Update annotations from annotator

Add simulated annotations for all 100 videos