Add support for multilevel/mixed effects models

[marcora]

The analysis of several experimental designs would benefit from the ability to use multilevel/mixed effects models.

The variancePartition package adds support for multilevel/mixed effects models to limma (under the name of dream) and I am wondering whether it could be added as an option to lipidr.

https://diseaseneurogenomics.github.io/variancePartition/

Hoffman GE, Roussos P (2020). “dream: Powerful differential expression analysis for repeated measures designs.” Bioinformatics. doi:10.1093/bioinformatics/btaa687