Determine whether sensitivity calculator synthesized beam is Gaussian

The sensitivity calculator reports a synthesized beam that is calculated by RASCIL. It is clear that in many cases this beam is not Gaussian and thus probably not useful for imaging and should not be used to calculate a surface brightness sensitivity. The calculator currently only assumes that a non-tapered Natural (or robust=2) image has a non-Gaussian beam, but this is far too crude and ideally each beam produced by RASCIL should be examined to see whether it is sufficiently close to a Gaussian. At the same time, bugs in RASCIL that were causing the synthesized beams to be miscalculated have now been fixed and so the time is right to investigate the beam properties in detail.

A start has been made on this. The attached plot shows a horizontal cut through the beam for five values of the robust parameter together with the Gaussian reported by RASCIL Also shown is the 2-D beam itself as in image. It is clear, for example, that with no tapering the beam is very non-Gaussian for robust values >= 0, but that it becomes Gaussian for various values of tapering. A metric (approximating chi-squared) is also shown that attempts to measure whether the RASCIL Gaussian is a good fit to the actual beam (for beam values > 0.1).

What is required is a metric that determines whether the beam is Gaussian or not and a way of making this known to the front-end. A suggestion is that the script that generates the weighting look-up table be modified to write out a beam image for each simulation, apply the metric to it and then write a column that signifies whether the beam was Gaussian or not.