[daip] Cal transfer & misc

[Athol Kemball]

Some misc. defects (while I still remember):

1) If the cal transfer writes multiple 1-D gain curves to
   the GC table, APCAL will reject all of them
   in GETRC (see NVALID > 1 check), leading to no cal.

2) The tick mark algorithm now fails in SNPLT for constant
   data (such as smoothed polarization offsets).

3) UVCOP appears not to rationalize the TY table for
   source selection, leading to many GETSOU warnings
   later for sources not found.