The q-space inplementation #18 offers a staggering N^2 speedup over the standard implementation.
However, some of the perturbations, such as IR and Raman, need to be re-implemented ad-hoc.
Using the prepare_perturbation_q, it should be easy.
Those that right now are missing are: