Is there any value in designing tests in such a way that they test multiple things at once while still being able to isolate which specific thing is responsible for breaking? I'm thinking of something like JMP.