ctecorr flexibility

[sbailey]

This PR makes the CTE correction code more flexible for cases when not correction is needed:

• when generating the CTE corrections, desi_fit_cte_night first checks if a CTE correction is even needed for a camera before complaining about things like BADCAMWORD being set. It also checks the science exposures and doesn't try to generate a CTE correction for a camera that is flagged as bad all night anyway.
  • @akremin this still only checks CAMWORD and BADCAMWORD; I could use some help for how to also check BADAMPS
• when using the CTE corrections, desi_preproc first checks if the correction is needed for a given camera before checking for the existence of $DESI_SPECTRO_REDUX/$SPECPROD/calibnight/NIGHT/ctecorr-NIGHT.csv . This restores the ability to preproc non-CTE-problematic cameras outside of the context of a production without having to specify --no-cte-corr. If DESI_SPECTRO_CALIB says that this camera on this night needs a CTE correction, then $DESI_SPECTRO_REDUX/$SPECPROD/calibnight/NIGHT/ctecorr-NIGHT.csv is still required unless the user opts-out with --no-cte-corr. @julienguy this should make you happier.

[akremin]

Overall this looks great. I have added some comments inline where I think one piece should be moved, and then I provide some additional functions that might be useful. I also address how to incorporate BADAMPS into this.

[sbailey]

@akremin I believe I have updated the camword logic to cover the cases you raised, including BADAMPS via erow_to_goodcamword. Please re-doublecheck my logic.

[akremin]

All of my requests were addressed in the latest commits. The implementation at line 604 in get_cte_images() is better than my original suggestion because it quickly skips nights where the camera isn't used, while also capturing the case I was concerned about.