Split out the unit tests into their own action. This decouples the unit tests
from the other tests. This means the gitlab schutzbot tests don't have to
wait until the unit tests are completed before they are triggered. This
also a step towards splitting out the tests so it is more obvious to see
which step is failing and why.