Using Explainable Artificial Intelligence To Improve Transparency Of Reinforcement Learning For Online Adaptive Flight Control