NLTK tokenizer, see https://github.com/nltk/nltk

We provide the nltk English tokenizer here for convenience. It can also be
downloaded using a python shell as follows:
```
import nltk
nltk.download('punkt')
```
