For now the tests only run the model once, but don't check any output. We should get some kind of test to work that actually checks results.