Compare language models with scored test suites.
Recent revisions are kept. Restoring one creates a new revision.