Rule change proposals

Generated by python scripts/annotation_kit.py proposals. One section per rule that the signal labels show misfiring, newest run overwrites the counts.

A false positive is a rule row labelled NOT_PRESENT (the match is not the capability at all) or BENIGN_CONTEXT (the capability is there but the context makes it low risk). NOT_PRESENT is the stronger signal that the pattern is wrong; BENIGN_CONTEXT usually means the pattern is right but needs context.

Counts come from labels. The two prose lines under each rule are written by the rater and are not generated: fill in why it misfires and what you would change. Anything still reading TODO has not been decided yet.

Do not edit a label file to make a rule look better. Proposals become issues and are applied in a later round, and the report says which round produced which numbers.

No false positives labelled yet. Label some signal rows, then rerun.