Modeling Bounded Rationality in Multi-Agent Simulations Using Rationally Inattentive Reinforcement Learning