The 291 visits include 39 Y, 34 u, 21 g, 12 r, 87 z, and 98 i -band. In total there are 291*61 = 17751 images.
I processed these cosmos data using stack w_2016_52 and the u/hfc/DM-8681 branch in obs_decam; ran pipe_drivers singleFrameDriver task with the reference catalog at /datasets/refcats/astrometry_net_data/2mass.20130404 using the slurm cluster. This produced 12050 calexps. The most common failure was "Unable to match sources" in astrometry (4133), but also got other errors such as too few matches, no viable psf candidates, and so on (I'll leave QA to future tickets as QA is not the goal of this ticket)
Ingest/butlerize using stack w_2016_51 with a hack in obs_decam ingestCalibs that ensures the "filter" column in the calibRegistry always be "NONE" for all bias masterCal data.