Question regarding rotations with observer-centric frames

I wish to rotate sky coordinates (`lon`, `lat`) between ecliptic, galactic and celestial coordinates. These are observercentric coordinates with origins given by a Cartesian heliocentric position `obspos` and a time of observation `obstime`. I am essentially trying to reproduce the rotations from healpy, with `healpy.rotator.Rotator` which rotates between `["E", "G", "C"]` with arbitrary origins (I think).

Using the `SkyCoord` object and the frames `BarycentricMeanEcliptic`, `Galactic` and `ICRS` seems to do the same as healpy's rotator. However, these are all solar system barycentric and it feels a bit weird to have users select these to represent their observercentric coordinates. Are there more fitting frames which I can use?

Ultimately my goal is to get heliocentric ecliptic unit vectors from the input coordinates and it feels like I have all the ingredients required to go directly from user a user specified frame `(E, G, C)` and `(obspos, obstime)`. I have tried supplying a `distance` or `location` parameter into the various astropy frames but I don't think i understand exactly how these are meant to be used.

This is from a suggestion by @lpsinger from the [pyOpenSci review](https://github.com/pyOpenSci/software-submission/issues/161) for the Astropy-affiliated [ZodiPy](https://github.com/Cosmoglobe/zodipy).

Start by reading the SkyCoord transformations involving BarycentricMeanEcliptic, Galactic, and ICRS, then compare their behavior with healpy.rotator.Rotator for E/G/C and the supplied obspos and obstime. Clarify with maintainers whether the desired outcome is a supported observer-centric transformation or documentation, with heliocentric ecliptic unit vectors as the acceptance criterion.

Written by the indexing model from the issue text.