When creating post stamp images, I have noticed that some detrened images have normal psfs flagged as cosmic rays. See figure CR_in_12_images.png (white areas are the areas flagged as CR). When creating the final image from stacking data this creates an nonphysical result in the variance image (see figure `sci_var_side_to_side` - science image on the left, variance on the right.)

Given that I have multiple images with same properties, I do not have to rely on the pipeline to detect CRs. I will recreate the poststamp images. This hints at some possible problems in the CR detection as well.