fix(bugbot-gate-selftest): pageInfo stripper loop visits every paged connection

[LukasWodka]

Summary

Fixes a vacuous case in scripts/tests/bugbot-gate-selftest.py: the pageInfo stripper loop only ever tested one of the two paged top-level connections. Tracked internally (Bugbot finding on the staging promotion #464, scripts/tests/bugbot-gate-selftest.py L913-L922 at that head).

Root cause, measured. The for name in gate.PAGED_TOPLEVEL loop ended in a break, so only checkSuites (the dict's first member) was exercised and reviewThreads never was. Its reviewThreads branch was also wrong on its own: QUERY.replace(" pageInfo { hasNextPage endCursor }\n", "", 1) uses a 6-space needle that is a substring of the 14-space-indented checkSuites line, which comes first in the query -- so even when reached it stripped checkSuites' pageInfo again and connections_missing_pageinfo answered ['checkSuites']. (Bugbot called it a no-op; measured, it is a wrong-target strip. Either way reviewThreads was never checked.) A pageInfo self-check that went blind for reviewThreads alone passed the suite green.

Change

• scripts/tests/bugbot-gate-selftest.py: the loop now mirrors the totalCount stripper directly above it --
  1. the two members are also written as literals next to the module's dict, so a member removed from PAGED_TOPLEVEL fails rather than shrinking the loop;
  2. the stripper is anchored on the connection's own name and is indentation-agnostic (re.subn), and its substitution count is asserted == 1;
  3. the detector must name exactly the stripped connection (== [name]) -- the neighbour, or both, is a wrong answer;
  4. the loop's visit list is compared to the literals afterwards, so a break reddens the suite.
     146 -> 151 assertions.
• scripts/tests/bugbot-gate-mutations.py: one new registry row, the exact regression the finding named -- connections_missing_pageinfo walking only the first connection (for name in list(PAGED_TOPLEVEL)[:1]:). The anchor is a two-line exact match; --dry resolves it once.

Test plan / evidence

Run with the CI Python (3.12 venv, pyyaml==6.0.2), from the worktree root.

• make check (ruff + shellcheck + house-rules + action-pins + mint-scope + reusable-no-cancel + lint-targets + actionlint + mutations-dry + all 31 selftests): green -- selftests-cover: all 31 selftests and 18 mutation runner(s) are wired to a target, and CI runs both tiers, bugbot-gate-selftest: 151 assertions, all passed, ==> check: green.
• make mutation-bugbot-gate (the CI shard for this suite): 64 mutation(s): 0 stale, 0 uncaught.
• make mutations-dry: every runner resolves, 0 stale.

Mutation proof, both directions:

• Registry row against the old suite (fix not yet applied): 64 mutation(s): 0 stale, 1 uncaught -- UNCAUGHT the pageInfo self-check walks only the first connection, so reviewThreads' cursor can be dropped from the query unnoticed. Against the fixed suite: caught ... by: dropping pageInfo from 'reviewThreads' is detected, and 'reviewThreads' alone is named -- detector said []. 0 uncaught.
• Reintroducing break after pageinfo_visited.append(name) in the fixed loop: suite exits 1 with FAIL: the pageInfo stripper loop visited every paged top-level connection -- visited ['checkSuites'], expected ['checkSuites', 'reviewThreads']. Restored.

No workflow or shell files touched; scripts/bugbot-gate.py itself is unchanged.

---

Note

Low Risk
Test-harness and mutation-registry changes only; production gate logic is untouched.

Overview
Closes a vacuous selftest hole where the pageInfo stripper only ever exercised checkSuites (early break) and never validated reviewThreads—so a gate self-check blind to reviewThreads could still pass.

bugbot-gate-selftest.py reworks that loop to mirror the existing totalCount stripper: literal checkSuites / reviewThreads membership checks, name-anchored re.subn with applied == 1, and connections_missing_pageinfo(stripped) == [name] so the detector must name exactly the connection stripped. A final assertion requires the loop to visit both connections (a reintroduced break fails).

bugbot-gate-mutations.py adds one mutation for the measured regression—connections_missing_pageinfo iterating only list(PAGED_TOPLEVEL)[:1]—so the old suite would have reported it uncaught.

scripts/bugbot-gate.py is unchanged; assertion count rises (146 → 151 per PR description).

^{Reviewed by Cursor Bugbot for commit 4d721b5. Bugbot is set up for automated code reviews on this repo. Configure here.}

[LukasWodka]

bugbot run

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 4d721b5. Configure here.}

[saqlainsyed007]

Approving. Verified against the source at 4d721b5: PAGED_TOPLEVEL has exactly checkSuites/reviewThreads, and the old break+6-space-substring needle really did leave reviewThreads unchecked (and mis-targeted checkSuites). The reworked loop is name-anchored and indentation-agnostic with applied == 1 and connections_missing_pageinfo(stripped) == [name], and the trailing sorted(pageinfo_visited) == sorted(PAGED_TOPLEVEL_LITERALS) reddens on a reintroduced break or an unlisted new member — mutation-proof both directions, and the new registry row's two-line anchor matches the live detector exactly. CI is fully green and there are no open review threads.