It would be really cool if one could store a document and its generated MinHash somewhere (like a database) and check for similarity with other documents at a later stage. As of now, this is not really possible to do, as one would have to regenerate the hash object with the document again since it's not possible to regenerate a MinHash instance from its digest or to get the jaccard distance to a hash in digest form.
I see an easy solution for both CSimHash and RSimHash: add a digest_jaccard method that takes a Vec<u32> digest and spits out the similarity. One could also consider modifying the existing jaccard methods instead, as only the digest of other is ever needed in them, so handing over the full MinHash object seems inefficient. Of course this would be a breaking change, maybe warranting a 0.3.0 release.
I'm happy to implement this, please let me know if this would be a welcome change and which option you'd prefer!