How to deal with various LSR definitions?

The wonderful new `SpectralCoord` PR (#10185) also introduced two new frames, `LSRK` and `LSRD`, and in hindsight I wonder if it is the right approach to define two new frames rather than, like for `GalactoCentric` allow the user to specify the assumed LSR velocity via a `ScienceState` - https://docs.astropy.org/en/latest/coordinates/galactocentric.html

This in part since I'm not sure our current definitions are entirely consistent: e.g., `LSRK` defines the velocity vector in `FK4` - isn't it strictly defined as co-moving with `FK4` then? It's quite likely that in practice this does not matter, but I worry a bit about mixing rough and fairly precise items in our transform graph.

Anyway, this mostly simply as a question, though one that we should answer before 4.1, hence the milestone. cc @astrofrog, @eteq, @adrn.

Start by reading PR #10185 and the GalactoCentric documentation linked in the issue, then compare how the LSRK and LSRD frames define their velocity vectors. Done means reaching and recording a decision on whether these should remain separate frames or use a configurable ScienceState before the 4.1 milestone.

Written by the indexing model from the issue text.