Investigate CR/bad column removal in the simulations

I've seen some single-pixel events leak into the extracted spectra, and was surprised to see them survive even in the 2-D images.  E.g. visit, 47 r1, (sky) fibreId=83,  x, y = 3509, 5302 ( ). I'd have expected a CR detection in the lambda detection to find this, and the extracted spectrum is a single high pixel too ( )