Sibyll*

[afedynitch]

This PR combines multiple items (sorry for that). The main novelty is SIBYLL* as shown at ICRC by F. Riehn:

1. Include SIBYLL* (thx @jncots)
2. When using Ninja for compilation, the FORTRAN files were pre-processed twice, once due to the -cpp flag in CMake, and a second time by Ninja itself that calls the preprocessor by default. Fixing this restores the quick build times on machines with many cores.
3. The allowed projectiles for all of the models have been reviews and the lists extended where appropriate
4. Some fixes to UrQMD file list (exclude the includes from f2py)

Opinions desired on:

1. If the SIBYLL* subclasses should be called using the underscore or camel case for consistency?
2. If I should put effort in using only one source file for SIBYLL2.3d and SIBYLL. In principle, one can restore original SIbyll2.3d behaviour from the sibyll source file.  

[jncots]

I think that:

• The name of classes should be changed standard camel style, like Sibyll23dStarNoEnh
• There should be only 1 file to avoid code duplication.

[HDembinski]

SIBYLL* is an unfortunate naming choice for us. I agree with @jncots that CapWords style for the new classes is the way to be PEP8 compliant, see https://peps.python.org/pep-0008/#class-names

[afedynitch]

@HDembinski I fixed the issues that you pointed out.

Further, I updated Sibyll* to patch release 2 that came in today with a bugfix. After your comments are resolved, I would like to merge this with to_release branch instead of main to bundle the numerous open PRs.

[afedynitch]

I implemented the comments by @jncots.

Also, I fixed the cross section readout for nuclei. It was necessary to include more categories into CrossSectionData to accommodate the special categories for nuclei, such as "production" and "quasielastic". I updated the tests to accomodate these changes. This will close #130 .