Review for NeurIPS paper: Model Class Reliance for Random Forests