Text this: Using deep reinforcement learning to design a tracking controller for a real-time hybrid simulation benchmark problem