METIS WCU FPmask with PSF convolution shows signal dropoff at the edge of the field

[oczoske]

The three figures show ImagePlanes for METIS-LMS observations (set_modes='wcu_lms') with the black-body source closed (wcu.set_bb_aperture=0) so that the radiation levels from the integrating sphere and the focal-plane mask are very similar. The case with no focal-plane mask (wcu.set_fpmask("open")) is as expected and desired. With the pinhole mask the bottom/left three slices have significantly lower signal than the rest; this indicates a problem in the across-slice ($\eta$) direction. Without psf convolution (metis['psf'].include=False), only the bottom/leftmost slice is affected.

The problem also arises in the along-slice direction ($\xi$):

The cause may be that the FP mask does not quite reach the edge of the slicer field (the green line stops one or two pixels short of the blue line). This drop is then broadened by the PSF convolution to give the orange line.

The same problem exists in wcu_img_lm mode, also on a few pixels at the edges of the (much wider) field.