There should be a few larger tests documenting and verifying real-world use cases. These should not be counted in code coverage. Depending on their run time, it might make sense to exclude these from PR runs.
[x] A data analysis case with data loaded from the disk
[x] Coupled Lorenz systems as in the Frenzel & Pompe paper (DOI: 10.1103/PhysRevLett.99.204101)
[x] Truncated distribution, with numerically evaluated MI.
There should be a few larger tests documenting and verifying real-world use cases. These should not be counted in code coverage. Depending on their run time, it might make sense to exclude these from PR runs.