Safe Reinforcement Learning via Hierarchical Adaptive Chance-Constraint Safeguards Written on March 6, 2024