[evlatests] Refraction model

[Barry Clark]

The current implementation of the on-line software uses a model atmosphere
that comes with calc.  I ran a few examples in simulation, to see the 
order of magnitude of differences one might see between this and the 
refraction as was implemented in the Modcomp on-line system (or almost).

The source was 3C286, and below are the delays calculated for an antenna
at location W36.

First line in each set is standard CALC, as we now use it, with a model.  
Second line is CALC without atmospheric model, as was used until September
4, 2007.  Third line uses the VLA delay calculation model, with
INFO: Temp=17.578947 Baro=780.799988 Dewpoint=-4.210526,
the conditions at the time I ran the simulation.

Delays are in picoseconds, elevations in radians.   

Only at the lowest elevation (ie, below 25d) is there a difference 
exceeding 1ps.  Under that particular set of conditions, which refraction
model is used is unlikely to make much difference. 
 
CALC= 2734123.9605070266  at  elev  1.3743722958993532
CALC= 2734122.845197123  at  elev  1.3743714236259774
CALC= 2734123.351990563  at  elev  1.3743714361559594
 
 
CALC= -9380242.737530498  at  elev  0.959893948667536
CALC= -9380239.072260308  at  elev  0.9598908799668739
CALC= -9380243.338049895  at  elev  0.9598909239769378
 
 
CALC= -11084612.665554537  at  elev  0.45364010866004795
CALC= -11084596.918855449  at  elev  0.45363114248983294
CALC= -11084613.76730326  at  elev  0.4536312693582801
 
 
CALC= -9562394.689062282  at  elev  0.21700425272557747
CALC= -9562340.012828124  at  elev  0.21698464468980933
CALC= -9562399.659676084  at  elev  0.21698490510831947