Python part still needs to be fixed, almost there, compiles in clang again

The key point is to speed up compilation, and make it easier to develop, by using the packaged C++ interface more extensively

No warnings from teqp code, all from externals

Can do binary mixture tracing now!

To allow for providing the mole fractions, in order to allow for the use with mixtures.

Closes #27

Finally done

So inconsistent in naming...

Switch TP VLE solver to Powell's dogleg method. Might be just a little bit more reliable, but hard to tell.

Add routine for TP flash for binary mixture, to solve for molar concentrations (and therefore mole fractions)

See https://github.com/usnistgov/teqp/issues/9

To avoid circular dependency in what is to come

More testing is still needed to identify other issues