Investigate Balanced Random Forest dependency

[potash]

We are currently using my branch of sklearn with the balanced feature I added to RandomForestClassifier. However, sklearn may never merge this feature. We want to be able to upgrade sklearn for bugfixes and compatibility with other packages. Two options I can think of:

1. Look into using BalancedBaggingClassifier from the imblearn package, which should be equivalent to BRF.

2. Refactor my implementation of BRF so that it can simply be an add-on to the existing sklearn codebase. Not sure how clean/easy this would be.