I don't think we need to go to town on this, but there are some tests where the test data has been generated by running the code under test, which isn't the best. If we can produce some independent sources of test data for some of these, that would be good.
But this shouldn't be a blocker for a V1 release if we don't get too much done on it!