Consider how to define mag_err for guide star selection

[taldcroft]

The current definition of mag_err for a star includes contributions from the catalog mag error plus per-readout error. In some cases for guide stars, e.g. comparing with nearby spoilers, it probably makes more sense to just consider the catalog mag error (i.e. the systematic error on mean mag, not per-readout mag).

This needs more thought but isn't required for the 4.0 release.

[jeanconn]

Sure, but it might not be a bad idea to get it reasonable now before we do more rigorous testing. I had pasted in mag_err for use in

1. the first cut of candidate stars (copying the mag_err < 1.0 filter from ACQ)
2. the per-stage magnitude filter
3. The error when comparing a star to its possible spoilers
4. The error when comparing a star to a possible column spoiler

I think 3 and 4 should revert to MAG_ACA_ERR. 2 is probably OK. 1 could go either way.