New cron job to automatically retry failed trigger jobs

[zackgalbreath]

Mitigates #1031

[zackgalbreath]

I initially tried to use the python-gitlab library (like we do for protected-publish), but frustratingly got stuck on actually retrying the trigger job. To retry a job you need to get() it first, but GitLab is funny about trigger jobs and returns 404 when you try to look them up individually.

[zackgalbreath]

This last force push reduced the amount of memory requested from 500M to 50M. This more closely matches what I'm seeing locally when running this command through time -l.