Decoding the Black Box: Extracting Explainable Decision Boundary Approximations from Machine Learning Models for Real Time Safety Assurance of the National Airspace