Add more accurate DAR models

[joshwalawender]

Added code to offer alternate DAR models. Based on DAR calibration work done with KPF, I have empirically shown that these DAR models are superior to the Filippenko 1982 equations. The two added DAR models (ERFA and slalib) are functionally the same, but the code path is different, only one of those options needs to be retained in principle.

The two new models differ from Filippenko 1982 at a small, but significant level. For example, at 60 degrees elevation, the difference in the DAR correction between the models peaks at 0.25 arcsec (or 1 slice on the small slicer) between 0.35 and 1.08 microns (the extremes of the KCWI wavelength range). At an elevation of 30, the difference increases to 0.77 arcsec over the KCWI wavelength range.

The ERFA model depends on importing the pyerfa package, while the slalib version adds code copied from the KAI repo. That code from KAI appears to be a python translation of the slalib fortran code from the Starlink software.

Minimal testing has been done so far. I've run both new models on some data I have on disk to demonstrate that the new DAR correction is not altering the results in an obviously non-physical manner.