v0.5.0: add supplement round (M13+M14, 310 ep across trial_1..7) after stripping observation.wrench to match A_modern schema

v0.5.0: add supplement round (M13+M14, 310 ep across trial_1..7) after stripping observation.wrench to match A_modern schema

v0.4.0: add worker_5_hr (trial_5 +43) and worker_jm_4 (trial_5/6 +100)

Dedup v3: drop worker_5_jm and worker_4_jm_v2 (all 121 trial_1 episodes were duplicates of worker_1_yj). trial_1: 271 → 141.

Re-merge with new uploads: worker_jm_3 (trial_5/6), worker_2_yj/3_jhlee/8_changyong (trial_2/3/8 fill); filters: 20Hz + dedup + success + schema A_modern.

Merge filtered episodes from collected_dataset (v4 12-scope, schema modern, 20Hz, dedup, success=True; ~1422 ep / 10 tasks)