faint-source proper motions

I returned to work on the faint-source proper motion project I started this summer at MPIA. Unfortunately, the fitting to real data is very fragile. There is quite a bit of work to do!