let the worker retry failed jobs

[hixi]

If a job is failed in certain conditions, retry.

Conditions:

• job started before local database is ready (error: has been starting up)
• job interrupted by redeployment
• other failures maybe included as well
• coarse granular is OK, so that every failure is retried, independently of what caused it:
  • Limit by max-retries.

Maybe new job status required: retrying.