[daip] Possible POSSM bug

[Walter Brisken]

Hi DAIP,

Craig and I stumbled upon a likely bug in POSSM last week.  We were 
plotting some VLBA data taken in 8 contiguous 32 MHz sub-bands produced 
by the new digital equipment.  SPFLG suggested that the phases matched 
perfectly at the sub-band edges, as one would expect.  In POSSM, phase 
jumps at each sub-band boundary were seen.

It turns out that the parameter SHIFT was unintentionally set to 0,1. 
Changing it to 0,0 produced the results we expected.  Craig correctly 
later stated that the shift's effect on phase should not care about the 
location of the sub-band edges.  I further tested this and for small 
shifts (SHIFT=0,0.01) do see continuous phase as a function of frequency. 
I suspect either there is an error in the handling of delay or there is 
some numerical precision issue.

Test data can be found on parallax, user id 2179, disk 4, catalog number 
81 (TD010.DIFX.6).