Review for NeurIPS paper: Security Analysis of Safe and Seldonian Reinforcement Learning Algorithms