Easier extension to other thermodynamic models for phase diagrams

[rkurchin]

Right now, ideal mixing is somewhat "hard coded" into the phase diagram construction code. One should ideally be able to feed in other thermodynamic models. I'm imagining something like:

• rename the current h and s functions to h_idealmix, s_idealmix
• make h and s keyword arguments to g_thermo
• use AD to go from g_thermo -> µ_thermo for any model

I'd want to do some testing for e.g. RK polynomials to check that using AD doesn't slow things down here too much; if so there should definitely be an option to feed in analytic derivatives for chemical potential since they exist for most models, which would really be the more time-consuming part of building this, since the first two bullet points above are pretty trivial to implement.