In the GarganText textflow, we need to send the text (i.e. title <> abstract) to a NLP ngrams postagging service.

If the abstract is not cleaned then it makes crash the NLP micro service, that is why I have not merged openalex yet.

It can be reproduced with the query "b12 AND children".

I have started but not finished to clean the text on the branch: clean-text-please