Explainable Reinforcement Learning In Flight Control Through Reward Decomposition