Anisotropic cosmic rays

[ibutsky]

This merge adds cosmic ray (CR) physics to the Dedner MHD solver (HydroMethod = 4) as well as anisotropic CR diffusion and streaming. 

The bulk of CR advection is solved in hydro_rk/(Riemann_LLF_MHD, Grid_MHDSourceTerms, and Grid_UpdateMHDPrim).  CR diffusion and streaming are calculated in Grid_ComputeAnisotropicDiffusion.C and Grid_ComputeCRStreaming.C

I'm still working adding tests for CR transport and improving the documentation, but I think this is ready to start the review process!

[jwise77]

We had a bug recently introduced that made the tests fail when you submitted your PR. It was fixed yesterday in #131. Could you merge the latest changes from enzo-project:master so the tests can run again?

[aemerick]

Apologies this took so long to get around to reviewing, but it overall looks good to me. Thanks for putting this in!!

Most of my comments (I think) are documentation / comment related, with a few asking for clarification on a couple things just because I wasn't sure what was going on. And a couple optimization suggests. Nothing major I hope.

Thanks again!!

(Also looks like merge conflict resolutions with the current enzo-project:master is needed)

[ibutsky]

Thanks @aemerick! I just implemented your suggestions.

[clairekope]

This PR has been open for a good long while and I'd like to see it be merged.
Everything in the code looks good to me. I would like to see annotated parameter files for the CR Shock Tube and CR Transport Test problem types in the run directory.

Andrew's requests have been addressed even though he is no longer available to give the "ok" in GitHub. Test suite fails because the latest Matplotlib has removed matplotlib._png as of 3.3.0. The docs fail to build on CircleCI but I was able to build them locally.

Note: Two parameters that were not added by Iryna, but were in the pre-existing CR
implementation, are not documented: CRmaxSoundSpeed and CosmologySimulationUniformCR.

All changes have been met

[gregbryan]

This looks very good! I made a number of suggestions that should be addressed, but once those are done (and they shouldn't be hard), this should be merged.

[clairekope]

Tested locally against main branch (2b20d1) sans tests using Hypre; not cleared to merge
test_results.txt

[clairekope]

Reverted Grid_InitializeUniformGrid.C; it had changes unrelated to CR functionality that just... broke Enzo! Tested push suite locally (without Hypre tests); cleared to merge