Currently the checks for new lint issues that flag builds as unstable/failed only compare to the last build for the current job.

This doesn't work well for a multibranch job that builds pull requests- the pull request job will generally always succeed because there is no baseline to compare the lint results to.

Instead, it would be nice to be able to specify a reference job. That way we could say "if the pull request has 10 new lint errors compared to the last build on develop, fail the build."