test: add coverage for reclaim event and JobExhaustedError

- reclaim event: verify it fires with the correct ids and a plausible
  stuckForMs when requeueStale reclaims a stale job; verify it is not
  emitted when nothing is stale
- JobExhaustedError on 'dead': verify a job at the attempt cap surfaces a
  JobExhaustedError with the correct queue/jobId/maxAttempts/attempts and
  that lastError reflects the prior attempt's recorded error