Allow for relative detector rotation in mosaic arrays

[oczoske]

Many instruments have several detectors in the detector focal plane. So far we have always assumed to be perfectly aligned among themselves and with the ImagePlane. In general, there will be small (or large: multiples of 90 degrees) relative rotations between them. Allowing for them in Scopesim requires another interpolating/resampling step from the ImagePlane onto the detectors. This should be done before any noise is added to the images, i.e. just before or after ExposureIntegration.

Related to #683