Unveiling Global Interactive Patterns across Graphs: Towards Interpretable Graph Neural Networks