Add more general NLTE solver

[chvogl]

We need to add a NLTE solver that includes additional physical processes (bound-free, two-photon decays etc.). Ideally we should also (1) take the non-linear nature of the equations into account (caused, for example, by the dependence of the Sobolev escape probabilities on the level populations) (2) treat the coupling to the ionization state directly (3) potentially couple to the thermal balance.