Support to yaha tokenizer

[keineahnung2345]

Hello, this is just an information to people who want to do NER.

I have found jieba tokenizer is not very good at tokenizing chinese surnames. For example: "我姓林" will be tokenized to "我" and "姓林". So I want to use yaha tokenizer instead.
And so far I had make my own yaha_tokenizer.py and conduct some corresponding change in registry.py. For people who also want to do NER, you can visit my repository:
https://github.com/keineahnung2345/Rasa_NLU_Chi
and find the two files.