Model sensor-to-sensor QE and PNRU variation

[cwwalter]

We may want to add sensor-to-sensor quantum efficiency (QE) variations to obs-lsst based on our measurements.

Pixel to pixel response non-uniformity (a combination of pixel size variation and pixel to pixel QE variation) could also be modeled but we know from work from @aaronroodman and all that this is a very small effect on the LSST sensors.

@aaronroodman : The paper refers to a LSST sensor. Is it a E2V prototype and can we assume the same will be true for ITL?