Additional test PSF diagnostics

[e-koch]

Other suggested metrics to include for the test imaging handler and report, building on #293 :

• beam axis ratio vs. major axis/robust value
• Comparison of Gaussian fits with different levels of cuts in the PSF:
  • Could be the direct 2D Gaussian fit varying the psf cutoff (e.g. psfcutoff in tclean, though the docs specify there's an algorithm to vary the cutoff in cases of failed fits: https://casadocs.readthedocs.io/en/stable/api/tt/casatasks.imaging.tclean.html#casatasks.imaging.tclean).
  • Or fitting an ellipse to the contour at different PSF levels (effectively how radio-beam solves the smallest common ellipse from sets of beams, but could be adapted for this use: https://github.com/radio-astro-tools/radio-beam/blob/main/radio_beam/commonbeam.py#L408)
• Options for 2D PSF model residuals rather than 1D radial profiles (which we currently use for the PSF metrics but are only well-suited when beams are close to circular)