Ship the Exit-50 Verdict Reading Without #2544's Headline-Sentence Arm? #2564

Question

#2544's acceptance has two arms. A round is reported as its own reading, and wait exits non-zero, where:

1. its verdict headline is not the clean one, or
2. its headline sentence names outstanding findings,

while its open-findings count reads zero and no thread or collapsed block is on the head.

#2562 implements the first arm as exit 50. It does not implement the second, a clean verdict whose headline sentence names findings. Copilot's second round on that pull request names "clean-verdict headline detection" as an issue in its headline, so the gap is on record there. Should the pull request ship without the second arm?

What the second arm would have to read

A clean verdict carries no structural marker that it names a finding, so the second arm can only be a reading of the prose. Over the second-format rounds measured on this repository, these clean-verdict headlines stated no open finding, listed none, and opened no thread:

• About nine named a remaining nit, such as "Only a minor prose nit remains, with no blocking issues."
• About twenty-five said nothing remained, such as "No unresolved review issues were identified."

Nothing in the body says whether a "remaining nit" is a new finding with no thread or an earlier round's thread that was already answered.

Options

1. Recommended: ship Read Copilot's Overview Sections Through One Recognizer in pr_review.py #2562 with the flagging-verdict arm only, and file the headline-sentence arm as its own issue. The flagging arm closes every instance of the shape pr_review.py Passes a Copilot Round Whose Headline Names Findings Beside a Zero Open-Findings Count #2544 was filed from, and it keys on the verdict, which is a structural marker. A prose reading of a clean verdict would exit non-zero on rounds that only mention an answered nit, and every wording change Copilot makes would need another reader change. Effect on the parked work: Read Copilot's Overview Sections Through One Recognizer in pr_review.py #2562 is ready apart from this answer, so it merges and promotes as it stands.
2. Implement a prose heuristic on Read Copilot's Overview Sections Through One Recognizer in pr_review.py #2562 now. A clean verdict whose headline names a "nit", a "finding", or "remains", with no counted finding, would also exit 50. This closes the blind spot. It also raises 50 on roughly a quarter of clean rounds, most of them already answered, and puts a wording-sensitive matcher into the gate. Effect on the parked work: another edit-and-review cycle on Read Copilot's Overview Sections Through One Recognizer in pr_review.py #2562 before it merges.
3. Drop the second arm from pr_review.py Passes a Copilot Round Whose Headline Names Findings Beside a Zero Open-Findings Count #2544 outright. The issue closes on the flagging arm alone, and no follow-up is filed. Effect on the parked work: the same as option 1, with nothing left open.

Belongs to

• Handoff Session Handoff [auto-2480]: Read Copilot's Collapsible Overview Blocks per #2480 #2547, track auto-2480.
• Blocks #2562, the feature to develop pull request. No promotion pull request is open for it yet.