Feature/collision refactor

[lazersos]

This pull requests moves the computation of various coulomb logarithms and Spitzer collisional time to a LIBSTELL based module so that both BEAMS3D and THRIFT can refernece a common database. The BEAMS3D regression tests all pass so there should be no chage BEASM3D performance.

@ajchcoelho will handle interfacing the new module into THRIFT.

[kudav]

Shouldn't we include the other coulomb logarithms and collision parameters (coll_op_nubeam) as well?

[lazersos]

Shouldn't we include the other coulomb logarithms and collision parameters (coll_op_nubeam) as well?

When a peer reviewed publication on NUBEAM exists where they properly describe the derivation of their operator and verify the behavior, then I think it's worth considering. For now it's in BEAMS3D to support the comparrision on ASDEX. The issue is that no device/diagnostic has enough fidelity to proivde a validation case to allow comparrision between different collisional models for fast ions. Additionally, this module is to support broader transport work and just happens to overlap a bit with BEAMS3D, hence the changes.

[kudav]

#319 has some bugfixes in commits 9163fc4 a7b8c7d and 1440758. Can we put them in here so they are fixed in the new version?

[lazersos]

@kudav If you have bugfixes, they should probably go in separate pull requests from the feature pull request. Create a branch based on develop and cherrypick each bugfix as a separate branch and make separate pull requests.

[kudav]

There was a bug in fact_crit which resulted in a critical velocity about two orders of magnitude too large because sqrt(electron_mass)^-1 was used instead of sqrt(plasma_mass)^-1. I discovered this when my slowing down run with the NRL calculation took only a minute (instead of an hour), by which time all particles had slowed down.
I also confirmed that fact_crit and fact_crit_weiland were the same save for the Z_plasma factor and removed the additional calculations. This also makes the difference between the two clear.