I wonder what is the current best practice in solving bulk-boundary coupled PDEs in FEniCS? For example: