edit changelog: the crawler shouldn't always retrieved all the feeds (espicially when there are a lot of feeds on the platform)
added tag to the API whitelist
various improvements to the crawler (better use of coroutines, test if an article should be updated). tags are now retrieved for the k-means clustering (previously achived with the content of articles)
Minor improvements to the layout template.
Some improvements for the manager and the asyncio crawler.
Fixed an error in the logging after the insertion of an article
Starting to improve the loging for Heroku (with logging.StreamHandler).
set result to None by default
Only set those two options when using SQLAlchemy.
updated python dependencies
Uses a strem handler for the logging on Heroku.
Rollback if the last query fails.