Improve critical property unit tests

[jonwzheng]

trafficstars

From #636, we noticed that some of the critical pressure estimates were wrong, despite the unit tests passing for them. Our unit tests did not sufficiently test against other implementations of Joback or values from the thesis. So the critical pressure was silently incorrect for a long time.

To prevent this from happening again, and ensure everything else is properly calculated, we should include checks against a diverse set of molecules, maybe from JRgui or hand-picked from Joback's papers.