Using `file_types` in JobArtifact model and `report_type` in Pipeline
as kwargs makes it easier to reason about.