[bananas] More calibration corrections

[Eric Greisen]

1. The revised for the 100'th time DOCAL issue still had trouble for
    CL table intervals less than or about 2 sec.  It was revised
    although there may be trouble less than 0.4 sec - which is
    awfully short for a CL table interval (maybe not an SN one).

2. DOPOL with SPECTRAL solutions was only allowed for the simple
    APPROX type of solution.  When DOPOL was true for other types
    (e.g. VLBI) the correct type might be lost after initialization
    and before finding values for the next time.  If that did not
    happen cal values = 0 were returned and all output data were
    0.  The other solution types are now allowed with SPECTRAL
    solutions although only the VLBI mode was tested.

A MNJ on 31DEC13 is recommended.  Note that 31DEC12 has the second
problem but it has not been corrected in that version.

Those of you bring optical images into AIPS may be interested in
a new task DSKEW to remove skew from such images.  It will be
available tomorrow on the MNJ.

Eric Greisen