We wanted to make it easy and robust to re-use fMRIprep confounds in nilearn, so we started a package called load_confounds. This package features common strategies as well as a flexible API to tweak any aspect of the denoising model.
The package is at an alpha dev stage, but we welcome feedback and contributions (we still need to implement scrubbing and ICA-AROMA). Also let us know if similar efforts exist which we may have missed.