Review for NeurIPS paper: Provably adaptive reinforcement learning in metric spaces