At the minute, we only support enrichment by global h-refinement. It would be good to consider other approaches.