Implementing MissForest imputation?

[acesnik]

Is your feature request related to a problem? Please describe.
This preprint from the Noble Lab shows great performance of the MissForest method, which differs from random forest. I'm wondering if that could be implemented here.

MissForest is written in R, and the peprint uses rpy2 to use it in an iPython Notebook. That said, there's this implementation in python, but it doesn't have any testing, but it could perhaps be a starting point, https://github.com/yuenshingyan/MissForest.

Describe the solution you'd like
Offering MissForest as an imputation option.

Describe alternatives you've considered
Using no imputation and the rpy2 setup -- this requires a bit of setup with getting the R packages in the right place and referencing them. The R package allows threading, but that isn't tested in the notebook above, which just uses one thread.