Review for NeurIPS paper: Independent Policy Gradient Methods for Competitive Reinforcement Learning