GitHub author grouping (#2022)

[daveoconnor]

This PR is related to ticket #2022.

Adds a new contribution app which records git commits, pr contributions, and issue contributions. The result ended up looking different to the design because email addresses were only available for commits.

For now there's a question around repos with multiple libraries. For commits we duplicate them which matches what the existing commit table does. For PR and Issue contributions they're on a per repo basis because there was no constraint issue there and no preexisting data.

With a github personal access token the management command takes up to 6 hours to run, which is 5 minutes of actual processing and 55 minutes of GraphQL rate limit sleep. That may be reduced with tokens which have higher access rates. This should be considered when setting up the celery task.

I've tried to optimize the calls as much as possible using Etags for commits and tweaking the GraphQL queries. Notes about this are documented with the queries.

Once the identity merging work is completed there'll need to be a change to allow users to claim identities.

This PR also fixes some issues with coverage tests.

[gregjkal]

Approved, with a handful of minor comments. I have some broader thoughts on efficiency, but will take those as future optimization opportunities.

[daveoconnor]

@gregjkal PR review responded to and adjusted - I'll leave this one to you to merge, it'll need to be run manually on the server anyway with ./manage.py update_contributions and then a task added at a later date.