Informed Reinforcement Learning for Situation-Aware Traffic Rule Exceptions