Index endpoints will automatically discover and use
the `to_eager_load` and `to_preload` declarations of their
representers. This takes away the possibility of forgetting
to add those to the scope.
I found a few endpoint that didn't use them, when other endpoints
rendering the same collection did.
Interestingly the SCM integration modules consistently declared
those methods but never used them.
* Receiving and Parsing GitHub webhooks
* added services for upserting PullRequests, GitHub users, and, checks.
* added the datamodels for storing these entities
* added logic to parse GitHub webhooks and save the entities above
* fix some specs, added model specs
* let UpsertPartialPullRequest return the upserted PR
* added the last missing specs
* avoid double-posts when editing pull requestsx
* attempt to reduce flakiness
* added pull requests API endpoint
* handle edge case of converting partial PRs to full PRs
* name raw input of an upsert service 'payload' to differentiate it from the prepared params
* handle edge case: partial upsert after full PR creation
* github_user does not reference users anymore
* fixed a doc comment
* use enums instead of custom validations
* WIP
* pairing with jens and some cleanup and test writing
* better migration, clear up old PRs with a cron job
* handle case of non-edit events which do not introduce new linked work packages, but still require a comment
Jan Sandbrink