Seperate the reading and the calculation of pixel positions

This makes it possible to know the nr of facets before knowing the image system (pixel scale, etc.)