[evlatests] L band tuning

[Ken Sowinski]

There is a disconnect between LoIfSetup in the executor and
the tuning algorithm in the OPT.  The executor enforces a
minimum L301 setting of 12928 MHz, so that any LO leak through
remains out of the downconverter input band.  But the OPT
allows it to tune as low as 11904 MHz when deciding which
Widar subband to use, leading to scripts which will not be
tunable at observe time.

If we set the OPT lower limit to 12928 MHz, and continue to
avoid using the zero'th subband the lowest we can tune in
L band is 1202 MHz.  This, I think, is unacceptable.  I have
changed the lower limit in the executor to one notch lower,
12672 MHz, and suggest that the OPT enforce the same limit.
This will allow a lowest L band tuning of 946 MHz which is
more than we need.

On the principle that the limit of 12928 MHz was set for a
reason, we should have a careful look at L band observations
to be sure that there are no adverse effects from LO leak
through.