v2.1.0-beta 104 languages, 10 tasks, dual backends

We are proud to announce the beta release of HanLP 2.1, which now offers 10 joint tasks on 104 languages: tokenization, lemmatization, part-of-speech tagging, token feature extraction, dependency parsing, constituency parsing, semantic role labeling, semantic dependency parsing, abstract meaning representation (AMR) parsing.