Graceful error handling when calling code host apis

[brendan-kellam]

The following adds graceful error handling for code host api calls. When a api call fails, an error message is raised and the config synchronization continues. Fixes #140

The side-effect of this is that if an api experiences some transient error (e.g., network issue, rate limits, etc.), then repositories will be marked stale when they actually are not. If autoDeleteStaleRepos is enabled, then the repo & it's index will be deleted.

Two things we can do come to mind:

1. We add a delay between when a repo is marked stale and when it is deleted. This way, it gives the config a change to be re-synced again (and the transient issues gone).
2. We add retry mechanisms to api calls s.t., we retry on known failures.