Benchmarking Chinese Commonsense Reasoning with a Multi-hop Reasoning Perspective