When running "compile"-mode exercises in interactive `verify` mode,
we print their output when we prompt the learner if they want to
continue. This improves the "experimentation" experience, since
trying different things does produce a visible change.
`watch` and `verify` do prompt the user to actively move to the
next exercise. This change fixes `run` to never prompt. Previously
it was inconsistent between "test" and "compile" exercises.
BREAKING CHANGE: we again change the behavior of the `run` command