Comment 1 for bug 65598

It would be cool if the test runner could be run in a mode where
it'll stop at the first failure and store the failure point. On
subsequent runs it should continue from this failure point,
repeating the same behaviour.