How to deploy a reinforcement learning agent in a simulation environment?