Add basic error-handling logic & retrying to API calls and web hooks

[I-Dont-Remember]

Initial version was a P.o.C. - everything worked, but is not very robust if we hit one-off 500s from Slack, or rate-limits, or the myriad of random junk that happens when you run a web service for anything more than toy interactions.

We should gracefully handle our own errors, as well as attempt to handle errors coming back from downstream clients like Slack and Webhooks.