regula_falsi solver does not respect regions of validity

[Yurlungur]

I discovered in #449 that the regula_falsi solver does not respect the bounds of the equation of state. This isn't a problem for equations of state that are robust against out-of-bounds calls. However several of the analytic equations of state are quite brittle. In those cases, it is better to use the original findRoot method I wrote. We may wish to make regula_falsi more robust at some point.

[jhp-lanl]

I think we often call it without bounds right? Maybe the overload with bounds should just respect the bounds and error out if they don't bound the solution.

[Yurlungur]

It needs bounds to initialize, as the algorithm is closer to bisection than Newton. But they appear not to be hard bounds. I wonder if we should add an optional argument that enforces the bounds.

[jhp-lanl]

Ah right. I don't know where I thought we didn't specify bounds.

Anyway, maybe we can add a flag to respect the bounds then and not try to expand them?