Reintroduce Dexter rhoV with grtrans mitigation

[bprather]

Rotativity solutions for transport should correctly recover the low-temperature (non-relativistic) solution for Faraday rotation. However, there are various ways of transitioning from a fit designed to be broadly applicable down to the low-T limit.

This exists because for the polarized comparison, I wanted to directly compare low-temperature rotativity behavior from various codes. grtrans, for example, generally uses the fits outlined in the grtrans paper -- however, it has a somewhat interesting 3-part smooth transition to the non-relativistic limit, which ipole did not originally capture when using these functions.

Since it now behaves well (and quite similarly overall to the default Shcherbakov fit), this PR re-introduces the Dexter rhoV fit as a runtime option (previously, it could not be enabled at all). Basically, I found it lying around in a testing branch and figured it might be a net benefit to have upstream as a reference or comparison option.