Review for NeurIPS paper: Sample Efficient Reinforcement Learning via Low-Rank Matrix Estimation