Create a Django management script that retrieves the latest alert summaries from a given time interval, cherry picks the most relevant alerts from each one & identifies the job ids which should be retriggered for each of these alerts. Basically, automate the Retrigger/backfill  steps that Code sheriffs currently do as described in https://bug1562178.bmoattachments.org/attachment.cgi?id=9074745
Do this by combining the algorithm of cherry picking alerts from alert summary with that of identifying jobs to retrigger. We'll be able to output the basic data needed to perform an actual retriggering operation.
The script must output a JSON at
stdout. This bug is more of a proof of concept. The code here will likely be revisited & adjusted, as we advance through the subtasks of bug 1562178.
Among the parameters of the script, we should be able to provide:
- time window (how much should the script look back; defaults to 1 hour)
- frameworks (talos, raptor, awsy etc)
- repositories (autoland, mozilla-inbound and/or mozilla-beta)