This ensures the project of an MR's head pipeline is eager loaded,
preventing an N+1 query problem from occurring when viewing the list of
MRs of a project.