Allow finalizers to decide whether they execute when the test is successful or failed