Distortion model is insufficiently flux-conserving in some cases

[mperrin]

The distortion model based on SIAF can in some cases result in PSFs with normalized intensity > 1:

fgs = webbpsf.FGS()
fgs.detector_position=(2040,2040)
psf_fgs = fgs.calc_psf(fov_pixels=101)
psf_fgs[0].data.sum(),psf_fgs[3].data.sum()

outputs:

(0.9894524802672955, 1.033390139745318)

In other words, the oversampled PSF is < 1.0 (as expected!), but the oversampled and distorted PSF now appears to have flux greater than unity. This is unphysical and must be incorrect.

It's not immediately clear how best to fix this: distortion is not strictly speaking flux conserving (since light can move in/out from the edges of the FOV). This is related to the imperfect approximation we use for this in webbpsf, applying the distortion as a model in the image plane rather than the pupil plane.

Simple workaround: check if it's > 1.0 and if so, limit the PSF intensity to 1.0?

This cases problems with some MIRAGE calculations, in particular, since MIRAGE checks the PSF grid is < 1.0 (after normalizing out by the oversampling**2, correctly)

FYI @Skyhawk172, @obi-wan76

[mperrin]

no action needed on this now in WebbPSF. Handled in MIRAGE PRs.