Bug? The PSF is not warped during image differencing

It looks like the template PSF is not warped in image differencing. See lines 407-411 of imagePsfMatch.py in ip_diffim. This leads to a much broader PSF being used for calculating the matching kernel if the template is at a finer plate scale than the science image. Also, if the PSF is elliptical, it may end up with the wrong rotation, and this may be causing an increase in "quadrupole" mis-subtractions.