Gym-μRTS: Toward Affordable Deep Reinforcement Learning Research in Real-Time Strategy Games