Does playermodel.latest_rows (the one door forecast.py:203, pwa.py:1709 and forecast_delivery.py:412 call) still select a played-but-uninvolved week, zeroing the 16 player_pbp_l4 features on the served board?

Owner lane: q-language

Question

Does playermodel.latest_rows (the one door forecast.py:203, pwa.py:1709 and forecast_delivery.py:412 call) still select a played-but-uninvolved week, zeroing the 16 player_pbp_l4 features on the served board?

Prediction

H-200 row: 40+ WR/TE rows stranded on the unrepaired board. Session prediction written after reading D-399 but before running: shipped arm 0 stranded; pre-D-399 arm (INVOLVED_WEEK=False) > 0.

Finding and verdict

CLOSED -- premise dead at HEAD. The defect H-200 names was fixed by D-399 in latest_rows and the invariant holds on the current served board (0/633), with a positive fixture showing it would catch 115 if the flag regressed.

Reasoning

Stops the row being re-drawn; confirms D-399's fix reaches all three board callers because they share latest_rows.

Next action

None in this row. The carries>targets downstream gate the row also named was not run: its upstream cause is 0 on the shipped board, so it could only measure other causes (not this row's question).