feat(24-05): CRUD write routes with pending/committed/failed audit lifecycle
- app/api/route53/zones/[zoneId]/records/route.ts: add POST (create)
- app/api/route53/zones/[zoneId]/records/[recordId]/route.ts: PATCH (update), DELETE
- All three write handlers: requireAdmin() first (D-04), validateRecordWrite() before
any AWS command (D-01), createPendingAuditLog() before submitRecordChange() (D-07/SC-3)
- Committed path: markAuditCommitted -> insertPulseCrudHistory (pulse_crud, SC-4) ->
mirror refresh; failed path: sanitizeAwsError -> markAuditFailed -> 502, no history row
- DELETE submits the exact mirror-read recordset (name/type/ttl/resourceRecords), never
client-supplied values, per Route 53's exact-match delete requirement
- recordId zone-prefix mismatch guard (T-24-17): 400 before any audit row or AWS call
- No staged-approval mechanism anywhere (D-03) — mutation executes on first request
- tsc clean; npm test 554/556 passing (2 pre-existing itglue-search failures, unrelated,
logged in deferred-items.md, already documented by plans 24-01/24-03)