Add Kosmo O'Neil priors to priors.py

[sblunt]

See https://arxiv.org/pdf/1809.05490.pdf, awesome work by K. Kosmo O'Neil. Would be great to have these priors as options in orbitize!

[sblunt]

A first-draft version of this feature for MCMC only is implemented in the obs-priors branch. Summary of game plan for OFTI incorporation that came out of discussion with Greg Martinez & Kelly Kosmo today:

• use rejection sampling to generate samples from the ObsPrior. Proposal distributions: log-uniform in P, uniform in e
• rotate Omega but don't scale sma

[sblunt]

The Jacobian currently implemented on the obs-priors branch isn't correct, since we're using a sma basis, and the Jacobian in Kelly's paper involves partial derivatives assuming we fit in a per basis. I need to calculate and implement the correct Jacobian.

[sblunt]

@isabelangelo I just merged the changes from main into my obs-priors branch, if you want to check out what I've coded up so far for this. As I understand it now, the remaining TODOs are:

• derive the analytical form for the standard orbitize basis
• update the code (potentially change the way I initially coded it to be consistent with the newer API)
• test & document
• (potentially): expand to other bases

Feel free to use the code I have as a starting point or to completely start over (I won't be offended either way).

[isabelangelo]

Thanks Sarah this is super helpful :) I'll look over it and see how much of the code we want to incorporate, but regardless it's good to have for reference!

[sblunt]

It's been 84 years, but finally this issue will be addressed in PR #365!

[sblunt]

Addressed!!!