Replica start up DB race conditions.

[jamshale]

If you start up multiple instances of the agent on a fresh DB with posgres storage the askar store will get a race condition with causes the agent to fail. It seems to create part of the DB initially but not the config table in time for the other instances.

This wouldn't really be much of an issue if the orchestration software retries the agent that failed. However, the agent should be hardened against this problem with a retry mechanism.