Optimizing the switching operation in monoclonal antibody production: Economic MPC and reinforcement learning