[daip] Position offsets in UVFIT

[Eric Greisen]

Andy Biggs wrote:
> Hi there. I'm trying to fit to the u,v visibilities of a source *far* from the phase centre with UVFIT. If I shift the data to the phase centre with UVFIX then the fitting goes fine with GPOS=0. However, if I try fitting to the un-UVFIX'd data, then I can't get sensible answers. I use IMDIST to tell me the x and y offsets of the source relative to the phase centre and use these in UVFIT - should that work?
> 
> I want to subtract the source from the data, taking into account its variation with HA due to the primary beam. I do this by model fitting as a function of time and don't want to UVFIX the data twice.

I have put back a corrected versin that is rather slower but uses 
correct maths for the phases and derivatives.  Sadly double precision 
sines and cosines are slow compared to linearized computations.
Hopefully, not too slow.

Eric Greisen