Review for NeurIPS paper: Breaking the Sample Size Barrier in Model-Based Reinforcement Learning with a Generative Model