Text this: Using Deep Reinforcement Learning to Decide Test Length