Handle redis outage gracefully

[lboeman]

On occasion, redis containers restart and during the time they are down some endpoints return a 500.
This can be seen for the list all users endpoint here:
https://sentry.io/organizations/solararbiter-rx/issues/2101968443/events/aa0660506dac45709fea2af03fd8dc49/
Probably requires special handling for certain cases.
For listing all users we won't be able to supply emails, for granting roles we won't be able to grant by email. We should either return partial data or an error code and message to try again.