[daip] Compressed continuum data has different weights

[Ralph Marson]

I have been comparing the output of FILLM to the aips++ equivalent
vlafiller. I have discovered that, for continuum data, the weights
generated by FILLM depend significantly on whether compressed data is
written. For example with compression I get:

    UVPRT1: Task UVPRT  (release of 31DEC01) begins
    UVPRT1: Finding the scaling parameters to set formats
     lucy      UVPRT(31DEC01)     24     16-MAR-2001  14:40:25    Page    1
    AK456       .L BAND.   1  Vol= 1  User=   24  Channel=    1  IF=  1
    Ref freq=  1.464899959 GHz  Ncor=  1  No. vis=    655356  Sort order= TB
    U, V, W are in Kilo wavelengths at the reference frequency
    Weights have been multiplied by      0.0001
     
    Source G980703  (    0)  RA   =  23 59  6.67  DEC  =  08 35  7.1
    Source= G980703     Freq=   1.464899959    Sort= TB        1   RR
        Time       Ant    U(Kilo )   V(Kilo )   W(Kilo )   Amp  Phas  Wt
     
     0/00:36:55.0  7-12      -3.64       3.52      -5.60   0.006  52  8.905
     0/00:36:55.0  6- 7       1.57      -1.52       2.41   0.004 131  9.172
     0/00:36:55.0  7-21      -2.18       2.11      -3.35   0.002 -21  9.508
     0/00:36:55.0  7- 8      -0.60      -1.99       0.87   0.002 -54 10.110

And without compression I get

     0/00:36:55.0  7-12      -3.64       3.52      -5.60   0.006  52  7.312
     0/00:36:55.0  6- 7       1.57      -1.52       2.41   0.004 131  8.179
     0/00:36:55.0  7-21      -2.18       2.11      -3.35   0.002 -21  8.171
     0/00:36:55.0  7- 8      -0.60      -1.99       0.87   0.002 -54  9.078

The aips++ vlafiller agrees with the weights in the uncompressed data so
I am hoping this has the correct values.

Cheers

Ralph