Expand typical methods such as DeepST, GraphST, stMMC.

[linjing-lab]

See modules for spatial domain, and now stMMC outperforms Leiden, SpaGCN, GraphST in ARI and NMI metrics. This issue hope typical methods such as DeepST, GraphST, stMMC can be integrated in dance, and utilize benchmark platform for more convenient experiments in modules import. For example, multi-modal contrastive learning framework such as conST can be new import in benchmark.