Session Handoff [auto-2187]: Pair Nonstandard details Markup Correctly When Blanking the Overview Narrative per #2187 #2382

Next steps, in priority order

1. Fix Pair Nonstandard details Markup Correctly When Blanking the Overview Narrative #2187 in scripts/pr_review.py's overview-narrative blanking (mask_narrative and the DETAILS_TAG pairing it relies on). Work each of the issue's three cases: a narrative closed with a nonstandard spelling such as </details >, a <details/> opener inside the narrative, and the report order of a label seen both inside and outside the narrative. Each case is either handled with a test or declined with observed-format evidence that rules it out, per the issue's done line. Any widened close pattern matches only where HTML actually ends the element: the reverted attempt in the issue body matched a mid-line </details x>, let its [^>]* tail swallow a following opener or a ### Suppressed comments heading, and accepted non-HTML whitespace through Python's \s, so do not repeat it. Leave the details_spans readers (suppressed_blocks, outside_diff_blocks, previously_missed_blocks) no worse. Add the test the issue's comment asks for: an unclosed narrative followed by - **Confidence:** high reports that label, and confirm it fails against a mutation that blanks an unclosed narrative to the end of the body. Build every fixture from constructed text. Done looks like: each case handled or declined with evidence, the full unit suite, ruff, mypy, the prose gate, the repo gate, and spec/validate.py pass, and a local-strict-review pass is recorded at the pushed head.
2. Drive the pull request into develop with Closes on promotion: #2187 in its body, scope develop.

External blockers

None known.

Internal dependencies

Step 2 follows step 1. #2372 (merged to develop for #2186, not yet promoted) changed details_spans in the same file, so base the work on current develop rather than main and build on that change.

State

No branch or worktree yet. Base the work on develop, per repo-worktree. The parked lane on track auto-2251 also targets scripts/pr_review.py and recorded no branch when it parked, so re-derive open pull requests and branches touching that file before starting.

The parked decision queue

None for this lane.

What the last round did

Nothing yet. The unattended picker created this lane from the open backlog.

What not to repeat

Nothing recorded yet beyond the reverted close pattern the issue body names.

New learnings

None yet.