Dev irnl correction

[JoschD]

Description

Adds the ability to correct the nonlinear errors in the IR via https://github.com/pylhc/irnl_rdt_correction instead of the fortran module. This is deactivated by default, but can be enabled via config.
There is a new example hllhc14_collision_python_irnl that uses the new correction.

Unresolved Problems:

• I added an optional pre-tuning before calculating the corrections, which corrects the orbit and tries to optimize the tune. I thought this would increase the accuracy of the correction, e.g. better beta-functions and better orbit for the calculation of the correction values, but instead, when this is activated, the tunes cannot be matched properly for the lhcb2 line after correction. This is very weird, as the correction values themselves do not look that different. By default this "pre-tuning" is therefore now deactivated.

• I am not sure why the correction in IR2 and IR8 is deactivated.

• xmask in general (probably unrelated to the PR)

  • I cannot get hllhc15_collision to work at all, as the acc-models-lhc path given in the config is outside of the repository
  • I cannot get 003_lumi_level.py to run in either hllhc14_collision_python_irnl or hllhc14_collision as it fails with "ValueError: On entry to DLASCL parameter number 4 had an illegal value" in the matching of the leveling.

Checklist

Mandatory:

• I have added tests to cover my changes
• All the tests are passing, including my new ones
• I described my changes in this PR description

Optional:

• The code I wrote follows good style practices (see PEP 8 and PEP 20).
• I have updated the docs in relation to my changes, if applicable
• I have tested also GPU contexts