check-upstream-firstlisted

# Upstream Fix Before Local Patch ## Overview A recurring, expensive failure mode: a framework symptom gets diagnosed from source, a patch gets designed, and the effort is wasted because the bug was **already fixed upstream** and the member is simply running an old release. The inverse waste also happens: an agent hand-edits a source tree that **nothing is running**, then reports the fix as applied. **Order of operations is the whole skill:** 1. Attribute the RUNNING process to a real install path. 🔴 **And before calling any member's install method a "regression", measure parity across ≥3 members.** A box differing from your memory is a hypothesis, not evidence. Twice in one session a member was wrongly reported as drifted when it was in fact the fleet standard (and the "fix" would have downgraded it). Corroborate with: does the running version even EXIST on PyPI? If not, pip cannot be the standard. Measure the install method on at least three machines before calling any single one a regression: one box differing from your memory is a hypothesis, not evidence. 2. Get the exact version + the latest upstream release. 3. Search upstream issues/PRs for the symptom. **Search OPEN PRs too, not just issues** — feature-gap prior art lives in PRs more often than issues, and an open-but-`dirty` PR may already contain your exact fix while being functionally abandoned. Check `mergeable_state`, age, and review-comment count before assuming a fix i