Add asynchronous article download and pre-processing

[abuzarmahmood]

This will allow us to directly load lemmatized and tokenized text when reading the json rather than processing it a the end