fix(gates): give review-gate the same report rows the guard and the queue read - #2875

fix(gates): give review-gate the same report rows the guard and the queue read

.github/scripts/review-gate.mjs built its reports rows in a third copy of a
shape that enqueue-guard.mjs:67 and merge-gate.mjs:224 had already been
fixed in (#2092, #2120). This copy drifted, and the report-only gate read a
different pull request than the queue gate does:

• no id — the handle a decision prints, and the one conclusionAt /
  stopRequestAt answer with;
• no review state — a review that asks for changes and carries no body is the
  whole of what a person writes there ("Request changes" is a button), and
  stopRequestAt (enqueue-rules.mjs) reads exactly that state as
  asked for changes;
• a review whose body is empty was dropped before it could be seen (A review asking for changes stops nothing: the review state is dropped and an empty body matches no stop #2119);
• a comment written on one line of the diff never reached the report, though it
  is a third place a person can ask not to merge (Read a stop request written on a review comment line #2091).

The effect: this report could answer PASS at a tip while the queue refused the
same tip for a standing request. The rows are now built once, by an exported
reportsFrom, so the drift is visible to a test instead of only to a diff.

What changed

• .github/scripts/review-gate.mjs — reportsFrom({ comments, reviews, reviewComments }) builds all three faces (comment, review with state, and
  kind: 'review-comment' for a diff-line comment whose body is not blank);
  main() uses it and passes only the conclusion faces to decide.
• The predicate isConclusionSource moved here from enqueue-rules.mjs, which
  re-exports it, so the guard and merge-gate.mjs keep their import. It is one
  definition, not a second copy: importing it the other way would have made an
  import cycle (enqueue-rules.mjs already imports parseReport from this
  file), and gate code should not have one.
• .github/scripts/review-gate.test.mjs — four tests on GitHub-API-shaped
  fixtures.

The three row builders stay in three places on purpose: collapsing them is a
separate change with its own equivalence test ("same fixture, three readers,
byte-identical output"), and merging it here would hide the drift instead of
pinning it.

Red first, then green

1. New tests against the base review-gate.mjs:
   SyntaxError: The requested module './review-gate.mjs' does not provide an export named 'isConclusionSource' → # tests 1 / # pass 0 / # fail 1,
   twice.
2. The fixed builder mutated back to the base shape (no id, no state, empty
   reviews filtered out, no review-comment rows — behaviour identical to base):
   • old assertions, twice: # tests 68 / # pass 68 / # fail 0;
   • new assertions, twice: # tests 72 / # pass 68 / # fail 4, exactly the
     four new tests, including
     + actual: 0 / - expected: 1 for the kept row and
     expected: 'asked for changes' for the stop request.
3. Builder restored: # tests 72 / # pass 72 / # fail 0, twice; the branch
   carries no throwaway edit. Neighbour suites enqueue-guard (155/155),
   merge-gate (52/52) and review-verdict (11/11) stay green with the
   re-export.

Verify

TZ=UTC npm run verify at this commit: core lint 0 errors, plugin lint 0
errors, 10 patches apply cleanly, core/ equals ds-bot/ + patches/ds-bot/,
canonical no-op; the .github + policy suite is 679 tests / 678 pass / 1 fail — the one red is the known #749 machine item, which names only this
host's .cache/wt-2547/** worktree and not any file of this change;
arch-docs: OK (3 changed file(s)).

No changelog fragment: gate plumbing, nothing a user sees.

Fixes #2870
Owner: Dev-5
Reviewer: Birfy
Written-by: Dev-5
Bot: Dev-5
Claim-Id: 9b6bdf22f96041bb8c08334b5d1f7310
Risk tier: T3 (judge: .github/scripts/enqueue-rules.mjs=T3, .github/scripts/review-gate.mjs=T3, .github/scripts/review-gate.test.mjs=T3)
Base: 0b77add

{
"pr": 2875,
"claim": "Dev-5",
"issue": 2870,
"summary": "review-gate built its reports rows in a third copy that had drifted from the guard and the queue gate: no id, no review state, empty-body reviews dropped, no diff-line comments. reportsFrom now builds all three faces once and the conclusion readers skip the review-comment kind through one predicate. A CHANGES_REQUESTED review with an empty body is now visible as a stop request instead of the report answering PASS. The three builders stay separate; collapsing them is its own change.",
"files": [
".github/scripts/review-gate.mjs",
".github/scripts/enqueue-rules.mjs",
".github/scripts/review-gate.test.mjs"
],
"checks": {
"ran": "TZ=UTC npm run verify; ARCH_DOCS_BASE=0b77addc07 npm --prefix core run arch-docs; review-gate/enqueue-guard/merge-gate/review-verdict suites",
"result": "green",
"at": "763e15078270c128dd5302f76a932c2eecf91e15"
},
"next": "waiting for a PASS at this tip (T3); the three-builder merge stays a separate issue"
}