p11 (5 Nov 04:20, before running): does 20CR-minus-fleet differ by region, and does removing that improve agreement?
Method: per-day anomaly a=SLP-20CR, binned 10x10 deg; region effect = median over ships of per-ship-median a in that cell (cells with >=5 ships); ships' a minus region effect -> region-adjusted cr; compare to offsets_v1 (unflagged, n>=30).
P1: region effects span >0.15 inHg between 10th and 90th pct cells.
P2: adjusted r >= 0.55 (was 0.40).
P3: Gorgon |diff| < 0.3 after adjustment (was 0.54).
Caveat: adjustment can remove real offsets where few ships share a cell.
