Integrate more NER datasets

[chen-yuxuan]

Datasets to be integrated in addition to CoNLL2003:

English datasets

• [x] Few-NERD
  • paper, repo
  • huggingface
  • names: supervised, inter, intra
  • splits: train, validation, test
  • features: ids, tokens, ner_tags, fine_ner_tags
• [x] OntoNotes 5.0
  • download
  • splits: train(115812), validation(15680), test(12217)
  • features: ids, tokens, pos_tags, parsing, ner_tags
• [x] WikiGold
  • paper
  • splits: test
  • features: ids, tokens, ner_tags
• [x] WikiAnn
  • huggingface
  • splits: train(20000), validation(10000), test(10000)
  • features: tokens, ner_tags, langs, spans
• [x] WNUT17
  • huggingface
  • splits: train(3394), validation(1009), test(1287)
  • features: ids, tokens, ner_tags
• [x] Lenovo Products (ask Leo)
  • features: ids, tokens, ner_tags, ner_bio_tags

German datasets

• [x] GermaNER
  • huggingface
  • splits: train (26200)
  • features: ids, tokens, ner_tags
• [x] Smartdata
  • repo
  • huggingface
  • features: ids, tokens, ner_tags
• [ ] Daystream
  • repo

[chen-yuxuan]

Remove branch extend_libraries because this has been already merged.