Thinking While Moving: Deep Reinforcement Learning in Concurrent Environments