[evlatests] Correction

[Ken Sowinski]

On Fri, 12 Feb 2010, Rick Perley wrote:

>    An important mistype in my previous report:  I wrote:
>
> 7) We have a sign problem somewhere -- applying the delay found by
> FRING, (through CLCOR) doubled the phase slope, instead of removing it.
> Eric remembers (and I agree) that this procedure used to work, and
> correctly handled both upper and lower sideband data.

I believe that changing the phase convention to get the
maps to be right way around also changed the delay convention.
Certainly the sense of the delays reported by telcal reversed
when MArtin changed the CBE.