HyPOLE: Hyperproperty-Guided Multi-Agent Reinforcement Learning under Partial Observation | AIChainDay