Automatically download NLTK resources

I keep forgetting and it's annoying. Can remove this from wiki, maybe 1st reference add something like:

```
import nltk

def ensure_resource(resource):
try:
nltk.data.find(resource)
except LookupError:
nltk.download(resource.split('/')[-1])

ensure_resource('taggers/averaged_perceptron_tagger_eng')
ensure_resource('tokenizers/punkt_tab')
```