Extract the fulltext fields for indexing

The json ES export does not contain the fulltext field at the moment.
This means that we can not search on existing binary text (the simple text part is indexed and accessible in the _all field).

Either we change the fulltextUpdater to save this info into ES.
Either we extend the core and dialect api to be able to retrieve the fulltext from the db.

Only the second option enable a fast initial import.

-> There is now a core API getBinaryFulltext() that return a mapping of binary fieldname/value