[FEAT] linked nodes with links (cf. `gargantext.constants` and `gargantext.models.Node.as_list()`
[FEAT] implemented an `ngrams` pseudo-attribute on `gargantext.models.Node` (available from REST api)
[FEAT] on REST API, renamed GET parameter `type[]` to `types[]`
[FEAT] `json_dumps()` can even serialize SQLAlchemy models and models now

(couldn't get it to work with 3.5)

[FEAT] serializing of JSON is much better and has its own module
[CODE] upgraded to Python 3.5 (`traceback.TracebackException`)
[CODE] did some cleaning, here and there...

(cf. commit 93957939)

[FEAT] implemented a kind of simple front-end REST library

[FEAT] deletions now cascade

[FEAT] started implementing REST API
(corpus view is still quite buggy, but without errors)

[CODE] improved bulk_insert rows management
[OPTI] ngrams extraction is treated little by little

(yeeehhhah!)

[CODE] removed unused imports

(documents are parsed from resources, but ngrams are not extracted yet)

 - immediate
 - threaded
 - managed with Celery
(see `gargantext.util.schedule`)

[FEAT] added some file managers (save, upload, download)
[FEAT] added Celery workers
[CODE] planned an interface for PubMed scrapping, see `gargantext.util.scrapping.pubmed`
[DOC] added a README and a TODO