Test results should be compared for all test

[lang-m]

For many tests we just test if the simulation runs. We should test the resulting values (with certain accuracy) to better detect changes/failures.