Incorporating Attributes and Multi-Scale Structures for Heterogeneous Graph Contrastive Learning