Implement ability to generate a physically reliable atmosphere model with an arbitrary composition

[adam-m-jcbs]

To increase the impact of dStar's science, we need atmosphere models with arbitrary composition. This, for example, can help it to couple with hydrodynamic stellar evolution codes like MESA or Kepler, providing a feedback loop that leads to more consistent, high-fidelity models of compact object phenomena.

This issue will track and aid in the implementation.