Distributionally Robust Model-based Reinforcement Learning with Large State Spaces