Gaussian model update

[nicholebarry]

I updated the gaussian source model generation. There are a couple of changes:

1) better gaussian parameter passing. I created a new structure so things are now much more compact and easier to manage.

2) including image plane -> uvw compression math from Jaiden's upcoming paper. Instead of calculating the mean compression from output from apply_astrometry, I'm calculating the expected compression using geometry. I've attached a screenshot of the relevant math.

3) switching the source_dft matrix ordering so that it's column major to be faster in IDL. This was a simple fix that significantly decreased walltime and mem requirements.

4) small bugs and general spacing.

The main branch (with Christene's LoBES catalog) took 381 min at 94 Gb for 69 118 sources. The new branch took 326 min at 65 Gb for the same sources. Running the main branch and the new branch with just point sources yielded the exact same power spectra.

Here is the difference between the old branch and the new branch (blue means less power in the new branch).

And here is the ratio difference between the old branch and the new branch (blue means more power was removed in the new branch.

The difference suggests that the geometrical argument behaves better than the mean calculation as projection effects increase towards the horizon.

[bhazelton]

Those power spectra look really promising!