STATUS-013: patched-in trees are a classify input every absorbed case must read right in status and commit right at post

Now
OPEN

RULING (gritzko, 2026-07-17): status/post read the patched-in (theirs) trees of in-scope wtlog patch rows as classify inputs so every patched-in case must be detected without staging (patch stages NOTHING). The known hole: a clean theirs-DELETE unlinks the file, nothing is staged, no stamp exists (no file to stamp) plain classify reads mis ("locally removed untracked") and post may not commit the removal. Umbrella: enumerate ALL patched-in cases, verify status renders and post commits each. The status-code vocabulary is codified in ONE table for get/patch/post/status (wiki page pending gritzko's approval); rmd is new (mod/del conflict, red like con).

Input

Context

classify.js already loads the DIS-057 patched-in stamps + theirs sha/mode for take-theirs ADDs (pStamp && t arm); STATUS-012 taught divergence to consult patch rows (ptch). Deletes and rmd have NO band slot (often no file), so they need the patch-row/theirs-tree consult, not stamps.

Goals

For each patched-in case: a repro asserting (a) the patch report row, (b) the between-patch-and-post jab status row (theirs-tree-aware, e.g. deleted del, never mis), (c) post commits the case correctly with no staged rows, (d) colors/order per the one status table.

Constraints

WIP

Design decisions

TODOs

Blockers and bummers

Outcome