Review for NeurIPS paper: A Local Temporal Difference Code for Distributional Reinforcement Learning