[FEATURE] Improve reliability against a crashed (and potentially restarting) TeraChem server

[coltonbh]

TeraChem crashes a lot. The server mode crashes when TeraChem does. I need logic in the client that handles that:

1. Some amount of retry time for connecting upon initial connect. Perhaps the server is being restarted by Docker
2. Correct error handling when a compute job crashes TeraChem. This means something like if the check_job_compelete function raises a ServerError I should return an AtomicResult with success=False and any other details I can provide (none at the moment).