Flaky tests

[jisantuc]

A lot of the interaction tests are flaky, for example:

• lots of 403s in this run that occur only intermittently
• random 404 in Issues >> RemoveLabel in this run

It's possible that these are pollution from simultaneous runs (i.e. -- since they're I think doing real life GitHub interaction, expectations about state sequences in one test run might be thwarted by another test run interacting with the same resources), in which case I'm not sure there's much that can be done without running against a test copy of the GitHub API (uhhh). Anyway, it would be good to know that that's the explanation vs. just speculating, and if that's not the explanation, to do something about the flakiness.