Modify donut code so that it can handle overlapping donuts

There are many cases where there are several overlapping lines. At the moment the donut code can only analyze simple images in which there are no overlaps of any kind. This greatly reduces the number of wavelengths that we can analyze.

In theory, this should be computationally ``easy''. I would assume that the donuts are the same, (i.e., wavefront aberrations are the same) and then there are only 3 extra parameters (ratio of intensity between two donuts, and dx and dy offset between two donuts.) Implementation could be a bit tricky...