Big changes in 2nd gen reference sims using diffuse maps

[jpober]

In 2.3_airy_flat_PS: ~Two of magnitude of difference in visibility amplitudes. (New runs are smaller.) Phases are different too; visibilities are not scaled copies of each other.

Change seems to appear between v1.2.1 and v1.2.2. (Although it may not account for all the difference.)

In 2.3_airy_GSM: Very strange change. Antennas seem to have three tiers of maximum autocorrelation amplitude in the new sims (as opposed to basically uniform autocorrelation amplitude).

See: https://www.overleaf.com/9613713963jthpjznmmyvx

[mkolopanis]

I remember in 7a3a127 we fixed a bug when nsky_parts > 1 had missing data. Have we re-run sims to compare since then?

[jpober]

Daniya did this testing and confirmed things were much better with EoR sky models. I think she also checked the reference simulations, but we are re-checking that now as well.