Enhancing Traffic Signal Control through Model-based Reinforcement Learning and Policy Reuse