Adapt the <only> syntax to branch@organisation/repo according to https://docs.gitlab.com/ee/ci/yaml/#onlyexcept-basic

Remove debugging ls. Activate double accuracy for the numerical tests. Deactivate 256 & 512 grid sizes.

Add gpu options to root cmake. Add the cmake option for passing the path to the numerial tests. Add git clone test to the CI.

Merge pull request 'Fix cuda::pow() bugfix, while passing doubles explicitly.' (#67) from peters/VirtualFluids_dev:bugfix/cuda_pow into develop

Reviewed-on: https://git.irmb.bau.tu-bs.de/VirtualFluids/VirtualFluids_dev/pulls/67

This pow-solution is not optimal but will work hopefully. The kernel performance will probably be decreased, because "double" calculations are 3-5 times slower compared to floating point operations.

According to the cuda documenation (https://docs.nvidia.com/cuda/cuda-math-api/group__CUDA__MATH__DOUBLE.html#group__CUDA__MATH__DOUBLE_1g6d36757715384dc18e0483aa1f04f6c7)
pow() gets two double arguments. With the newest VisualStudio Version 16.8 these arguments needs to be passed as double explicitly.

Set jobs to manual, as long the gitz gitlab has no windows runner and sonarqube is not transferred to one of the runners.