[Fooftilly]

Summary

Incremental B1 expansion under #651, building on merged Vitest coverage pilot #664.

Extend the existing .github/workflows/vitest-coverage-pilot.yml job to also measure frontend-app/src/query/retry.ts with its existing retry.test.ts, alongside src/lifecycle/tab-leave.ts and tab-leave.test.ts.

retry.ts controls whether safe-GET API requests retry after network errors, 502/503/504 responses, aborts, and exhausted attempts, plus backoff timing. It is a small correctness-sensitive module whose branch coverage is meaningful.

Deliberately narrow scope

• One existing workflow file, no product/runtime/frontend test changes and no new jobs or dependencies. Reuse the already pinned @vitest/coverage-v8@5.0.2 and existing npm ci --ignore-scripts.
• Run only two existing Vitest test files in one process and collect line + branch data from only two source files.
• Extend the job summary to show per-module numerators/denominators and percentages, plus a clearly labeled two-module aggregate, not a misleading repo-wide percentage.
• Fail the job if either requested source file is absent, another source file appears, either metric has a zero denominator, or totals do not reconcile. A low percentage never fails it.
• Trigger the pilot when the workflow or either source/test module changes, plus the existing config/manifest triggers.
• No coverage thresholds, artifacts, external uploads, new permissions, required-check renames, or changes to the ordinary Vue / PRKS test gates.

The existing job ID/name remains stable even though it now reports on two modules, to avoid changing check identities; the GitHub Actions summary explicitly names both scopes.

First verified baseline

Vitest coverage run 38581099227 succeeded on head 406d073 with 15/15 tests passed.

Measured source module	Lines	Branches
src/lifecycle/tab-leave.ts	77/94 (81.9%)	53/84 (63.1%)
src/query/retry.ts	7/7 (100.0%)	6/6 (100.0%)
Two-module aggregate only	84/101 (83.2%)	59/90 (65.5%)

This is strictly a two-module measurement, not global frontend coverage. See other CI and review results before merging.

Acceptance checks

• CI must confirm the grouped inclusion glob measures exactly src/lifecycle/tab-leave.ts and src/query/retry.ts. Inspect the job-summary per-module line/branch totals, not only overall success.
• The 12 pre-existing tab-leave tests and 3 pre-existing retry tests must pass.
• Verify regular Static Analysis (including Vue), PRKS Test Gate, actionlint, zizmor, CodeQL, and other triggered checks; check any reviewer findings.
• Previous ci: pilot informational Vitest coverage for tab-leave decisions (#651) #664 baseline for tab-leave.ts: 77/94 lines (81.9%) and 53/84 branches (63.1%); preserve or investigate any difference. Record observed retry.ts baseline from CI before merge.

Base: master@befacaad75bf6bb00fc72da6a5f4b818cf3b1b19.

Related: #651 (B1), #662, #664. No merge without maintainer approval.