at the moment the tests for codecov estimate the coverage of some dummy code base
but it'd be nice to know the coverage of codecov itself that is run during the tests: the problem I ran into is that we get some infinite recursion when trying to do that