Warnings can come in various severity levels. We want to least be able to check if a warning message is contained  in the output, irrespective of its severity level (this should already be possible?). We might also want to verify that a warning was emitted with a given severity level.