Make critical point calculation more robust

[g-bauer]

Depending on initial values, the critical point routine may return the wrong state. We currently only check if the result for a temperature is OK (and then we don't check additional temperatures) but a more rigorous check is needed.

[ianhbell]

Are you concerned with pure fluids or mixtures here?

[g-bauer]

@ianhbell Pure fluids for now. I noticed in the SAFT-VR Mie notebook when I tried to reproduce Lafitte's results that some seemingly converged results are not correct. Changing the initial temperature solves this and so does a stability analysis - but is expensive to run.

[ianhbell]

Also, as I gather, they used something like centered finite derivatives in their thermodynamic modeling for higher derivatives, so you should not expect agreement to be as good as with FeOs, teqp, or Clapeyron.jl. I couldn't reproduce their critical points in their paper to all their given digits.