when running tests in parallel I had failures on a few tests because they were stepping on each other.
The correct fix would be to change the tests so that they do not clash, but serializing the execution is a valid workaround until somebody wants to correctly tackle the problem.
