Factsumm

- resolve SentencePiece dependency issue

- remove Flair, we only use huggingface for named entity recognition
- remove OpenIE-based `open-scheme` triple score (too unstable)
- apply hf's official named entity recognition aggregation strategy
- fix BERTScore length issues by applying iteration with summary lines

Update BERTScore based Module (See Sec 4.1 from https://arxiv.org/pdf/2005.03754.pdf)

python
>>> factsumm = FactSumm()
>>> factsumm.calculate_bert_score(article, summary)
BERTScore Score