Some turbulence models are not working correctly

[marcomangano]

Description

Opening this issue to keep track of the conversation and work around non-default turbulence models in ADflow.
This is mainly to keep track of the work done by @DGCaprace in #107, as we decided to merge the PR in the current state and continue the development work in the new turb-bugfix branch.

Related issues and PRs:

• Bugfix on turbulence models #107 addresses Menter SST turbulence model outputs NaNs #99 but there are other fixes still pending - check the PR discussion for more details.
• SA-R implementation is incorrect #156 also falls under this bigger issue

NOTE: Any PR merged into turb-bugfix should not be squashed, so to keep better track of the development work. Only the final merge of turb-bugfix into main (whenever it will happen) should be squashed.

Steps to reproduce issue

See related issues

Current behavior

See related issues

Expected behavior

All turbulence models implemented in ADflow should work correctly without major convergence issues, segfaults, or NaN