I've attached the command log and the verification notebooks. There are two notebooks for each output product. For the initial bias and dark runs, the parallel overscan was not enabled during verification (not during generation), and was corrected in the second run. The flats had residual dust rings due to the dust pattern differing between 2023-04-25 and 2023-04-27. The 2023-04-27 data was excluded from the second generation run, as the 2023-04-25 pattern appears to be more common.