No cosmics and spurious satellite masks in some 90prime V2.3 masks

[djschlegel]

While digging into ticket #644, have found that 90prime masks also have some unintended changes going from the versions used for DR8 and DR9.

For the OOD masks for 90prime in use for DR9 (V2.3), there are typically no DQ=5 cosmics included in the masks. Here's a list of arbitrary exposures with no cosmics in CCD3: 90prime/CP/V2.3/CP20170222/ksb_170223_045941_ood_g_ls9.fits.fz 90prime/CP/V2.3/CP20170319/ksb_170320_040334_ood_r_ls9.fits.fz 90prime/CP/V2.3/CP20170426/ksb_170427_052652_ood_g_ls9.fits.fz

Spurious long streaks (DQ=8) occasionally appear in the masks, but not as often as for mosaic. Here's a list of arbitrary exposures with long streaks in CCD3: 90prime/CP/V2.3/CP20160209/ksb_160210_042208_ood_g_ls9.fits.fz 90prime/CP/V2.3/CP20170220/ksb_170221_083844_ood_r_ls9.fits.fz 90prime/CP/V2.3/CP20170520/ksb_170521_100354_ood_g_ls9.fits.fz

[djschlegel]

Frank reminds us that the CR-finding is turned off when the seeing is very good on the 90prime images, since those images are under-sampled. Looking at one particular night (90prime/V2.3/CP20170425), all images with SEEINGP < 3.6 pix have CR-finding disabled, all images with SEEINGP > 4.2 pix have CR-finding enabled, and that images in-between may or may not have CR-finding enabled. In other words, I don't see a hard cutoff for when this is enabled or not.

[fvaldes]

The CR cutoff is seeingp < 4. However, each CCD can have a different seeing value so it may not be uniform across the exposure.

[fvaldes]

The same fix as for Mosaic noted in #644 worked for the example noted earlier: ksb_160210_042208_ood_g_ls9.fits.fz.

So #644 and #645 are the same and have been fixed for future processing.

[moustakas]

Looks like this issue was addressed; closing.