Robust tests of Work Queue functionality

[RyanGutenkunst]

I recently discovered that InferDM using Work Queue was broken by the new bestfit_p0 functionality. (Fixed here: #38 ) We need robust tests to ensure this does not happen in the future.

I see there are some Work Queue tests in the test suite, but they are skipped @pytest.mark.skip(). Why are these tests being skipped? If run length is an issue, they can be made to run very quickly by setting --maxeval very small.