Implement retry with exponential backoff on all gcloud service calls

[nhammond]

Google recommends doing this for all 5xx and 429 errors.
https://cloud.google.com/storage/docs/exponential-backoff

[nhammond]

Consider also handling "socket.error: [Errno 104] Connection reset by peer", as seen in #361

[nhammond]

Also this: ssl.SSLError: ('The read operation timed out',)

#364

[nhammond]

Retries added for everything in filemanager and task_monitor. There may be more work to do in the playbooks. 1a2ada9

[nhammond]

I searched playbooks and found retries on everything except Docker pull commands. Added retries to docker pull in 7dc5b79.

Retries in ansible playbooks are still fixed interval, since ansible doesn't yet enable exponential backoff. #388 is open for changing to exponential backoff when avaialble.