With `gitlab-experiment` 0.5.2 a bug got fixed where stubbing
experiments in tests didn't work. With that fix, we can get rid of
overriding the `should_track?` method to always return true, since
we can now rely on the actual implemention, where it can exclude
actors.