Collimator feature (b-camera) on 20240223 sframesky images

[araichoor]

on 20240223, the collimator feature in the b-camera of the sframesky images are clearly visible (I didn t see any obvious effect on the redshifts, so I ve QA-validated the tiles): https://data.desi.lbl.gov/desi/spectro/redux/daily/nightqa/20240223/sframesky-20240223.pdf.

e.g. for expid=226845:

the amplitude of the effect varies through the exposures, but is overall quite high.
also I notice that the effect is strong for 226840, but almost disappears on the next exposure 226841 (and then comes back); I report just in case.

gathering here elements from the slack discussion (https://desisurvey.slack.com/archives/C01HNN87Y7J/p1708883336494629 and after):

• @djschlegel : "I’m concerned that we might not be running with the correct calibration vectors and are seeing the collimator absorption dips."
• @akremin : "I can confirm that the calibrations were processed first, all dependencies were followed, and the calibrations were used correctly by the science exposure scripts; Looking at the humidities, there was a shift from when the science exposures were taken with ~25% humidity and when the cals were taken with ~13% humidity. But I don’t think that level of variation is unheard of or should cause these issues."

[schlafly]

@julienguy , sorry, I think you're the best person to debug this and to declare if we're fine or if something's up with the fiberflat humidity correction. It does look especially poor on this night. Sky level was ~13---high but not exceptionally so.