[evlatests] UVW fixed; new archive revision number

[Eric Greisen]

Ed Fomalont writes:
 > Hi Walter,
 > 
 >     Thanks for the fix.  A related three part question.  People often 
 > run UVFIX anyway (not for the sign problem over the last few days) to 1) 
 > correct for the old VLA time error of 0.5 sample; 2) increase the 
 > accuracy of the u,v,w that are computed by the on-line system; 3) 
 > correct for the field stretching associated with annual aberration.  Are 
 > any or all of these previous minor 'errors' now corrected with the 
 > on-line uvw and times by mod-comp free, or should UVFIX be run for the 
 > wide-field high-precision work anyway?
 > 
 >     Ed

The format still has the end time - but there has been no reason for
FILLM to regard that as gospel.  Do you have any idea why we have kept
that absurdity for so long?  I am about to put back a new FILLM
that corrects the time by half the integration time.  All data of any
obs date will be correct henceforth.  The issue is documenting this
change.  I don't know about reason 2 even with ModComps...

Eric