sigmas_lnf for radial velocities with system velocity, and request for a simpler "jitter" term

[jsinkbaek]

Hello!

This is a question about the fractional increase of uncertainties through the "sigmas_lnf" parameter.

For radial velocity data, it seems to me that it scales based on the full radial velocity model; meaning that if the system radial velocity is not 0 km/s, the uncertainty will be applied asymmetrically to the RV curve.

Is that correct? I've looked at the code on https://github.com/phoebe-project/phoebe2/blob/7bc4489a5b2bd3f349291deec850589537109c0b/phoebe/parameters/parameters.py#LL3858C18-L3858C18
I cannot find any clear indication that the system RV is "ignored" when performing this rescaling.

Separately, I have a feature request. Some exoplanet people forego the use of a fractional RV uncertainty in favor of a less conservative additive component, simply a jitter term that can be sampled over as a constant quadrature addition to all RV uncertainties:

sigma**2 = formal**2 + jitter**2

Would you be interested in implementing that feature into PHOEBE? This would only work in cases where the additional "uncertainty" can be assumed to be independent from the model.

[kecnry]

The first part of this should be covered by #901 and I think the additional request for a jitter term might be possible through the possible upcoming support for custom dataset features in #902 or using the existing functionality for custom cost function if you want to give either of those a shot and see if they are flexible enough for what you need.