feat(autopilot): fleet-overview-filters — default-actionable fleet filters + 27-locale i18n

[pacphi]

Ready for human review — do NOT auto-merge

Autopilot integration PR for the fleet-overview-filters feature. Delivered across develop; all phase gates passed with green CI. A human decides whether to merge into main.

What shipped

The Remediation Fleet Overview now defaults to the rows operators can act on, via two independent, combinable client-side toggle filters — with a distinct empty state when the filters hide everything.

• FleetOverview.tsx: onlyWithPrs (openPrCount > 0) + onlyEligible (eligible === true) boolean states, both default ON, AND-combined into a derived filteredFleet over the already-fetched useFleetRemediation() rows. No backend/API change.
• Two shadcn Switch + Label toggles above the table (mirrors PolicyEditor/RepositoryFilterSettings); toggling either widens the view.
• Distinct filtered-empty branch (remediation:fleet.filters.noMatch) with a "Show all repositories" escape + always-visible toggles — separate from the existing zero-repos remediation:fleet.empty state.
• Full 27-locale i18n for the new remediation:fleet.filters.* keys (onlyWithPrs, onlyEligible, noMatch, clear, groupLabel); frontend/src/i18n/types.ts regenerated in sync.
• A11y: role="group" + aria-label on the filter cluster, role="status" aria-live="polite" on the no-match state.

Phase gate summary

Phase	Scope	Gate	Marker
0	Filters + filtered-empty state + vitest specs + full 27-locale i18n	✅ PASSED — all 21 CI checks green	9ea5c91 (#75)

The plan was originally 2 phases (filters+EN, then translate); Ampel CI's i18n missing gate fails any PR with en-only keys, so the 27-locale translation was carried in phase 0's PR and the plan was collapsed to one phase to match what shipped.

Verification (on develop)

• make test-frontend — 924 passed, 6 skipped (61 files); 13 FleetOverview specs (each filter proven independently, toggle round-trip, single-toggle escape from empty, revealed-row preview).
• make lint-frontend · make format-check · type-check — clean.
• i18n missing clean · coverage 100% · types.ts in sync · all 27 locales carry the 5 keys.

Review signals

• qe-code-reviewer: no correctness bugs; empty-state distinction airtight; clear-escape sound; both filters proven independently. 6 LOW/INFO findings all addressed.
• /code-review (high, workflow): 2 confirmed findings, both working-as-designed per spec — (1) both filters default ON, so a quiet-period fleet (all eligible, zero open PRs) lands on the filtered-empty state on first load. ⚠️ This is a deliberate UX tradeoff — flag if product intent differs. (2) new-keys-only-in-en was resolved by translating all locales in the same PR.
• Post-translate audit caught and repaired domain-noun mistranslations (repos→pawnshop/repurchase/cure, PR→public relations) by hardening the EN source to full words.

🤖 Automated autopilot feature. Ready for human review; do NOT auto-merge.

[pacphi]

✅ Snyk checks have passed. No issues have been found so far.

Status	Scan Engine	Critical	High	Medium	Low	Total (0)
✅	Open Source Security	0	0	0	0	0 issues

💻 Catch issues earlier using the plugins for VS Code, JetBrains IDEs, Visual Studio, and Eclipse.

[github-actions]

📊 Coverage Report

Component	Coverage	Status
Backend	N/A%	❓
Frontend	N/A%	❓
Overall	N/A%	❓

Coverage Thresholds

• 🟢 Green: ≥ 80% (target)
• 🟡 Yellow: 60-79% (acceptable)
• 🔴 Red: < 60% (needs improvement)

---

Coverage reports generated by CI workflow