Support variable backoff on retry

[benbuckman]

When a job fails currently, it has a backoff of 2^count seconds.

https://github.com/Shyp/rickover/blob/dba4e9add9ec4c0541bb4a777fa8d209feeb9221/services/status_callback.go#L88

Sometimes it's useful to wait much longer, like 30 seconds between runs. It would be nice if the base (2) were variable per job in the database. For example, setting it to 5 would make the backoff be 5s, 25s, 125s, etc. (Maybe the ^count exponent should also be variable? But that would be harder to implement and probably not necessary.)