Simulate prism observations

[rmjarvis]

Rebekah Hounsell wants to use GalSim to simulate prism observations from Roman. This mostly involves implementing an old (and closed because part of it was done) issue of 1-d degenerate objects. In this case, the 1-d function would be the SED in the direction of the prism dispersion.

[matroxel]

Working on this with SN PIT. First prototype prism model implemented as photon op in https://github.com/matroxel/roman_imsim/blob/prism/roman_imsim/photonOps.py

Need to update some roman module things to enable prism elements that we had previously never implemented, which will happen in this branch.

[arunkannawadi]

I think I'll take this issue over from Troxel. I have computed sky backgrounds corresponding to prism and grism, but noticing that I don't get the current values for the other filters exactly. I know they are accurate to only 10% or so, but I was hoping to get the exact values reproduced by existing filters as a cross check. I'll raise a PR later today so we can see if they seem reasonable.

[arunkannawadi]

Troxel tells me that @rmandelb is integrating the code to compute the sky backgrounds on the fly instead of using a lookup table. That's great!

But in the mean time, could we not use the sky backgrounds for SNPrism that are already here? I see that it was added in #1017 however, the sky level for prism and grism bands continue to be skipped here: https://github.com/GalSim-developers/GalSim/blob/releases/2.5/galsim/roman/roman_bandpass.py#L138-L139

Can we remove those lines?

[rmandelb]

Hi @arunkannawadi - For now, for various practical reasons, I have integrated the code to compute sky backgrounds on the fly (based on Troxel's port of Chris's ETC code, with both of their permission) into a separate repo in which we are developing tools for Euclid-like simulations. However, the idea is to make that public soon. If we then want to move that out of the Euclid-like code base and into GalSim, that would be fine.