Text this: A Reinforcement Learning Engine With Reduced Action and State Space for Scalable Cyber‐Physical Optimal Response