DocExtract: several improvements

• Ability to load journals from journal records database

• Moves html generation of /textmining to a dedicated template file so that it can be overwritten by INPSIRE

• Reworks the removal of [] and () around report numbers
• Which fixes a bug that would add extra characters in the misc string
• Only unidecode strings when authors are present

• Replaced the [A-Z] with a list of uppercase characters that is computed at load.

• Adds handling for caret in pdf2text results

• We can know check for existing tickets for a given recid. Which means we can always create tickets for extracted references by making sure there is not existing ticket.

• Adds a function to know if we can safely extract references from a record.

• Submits a single bibupload task, for a refextract task run instead of one per record.

Signed-off-by: Alessio Deiana <alessio.deiana@cern.ch>
Tested-by: Samuele Kaplun <samuele.kaplun@cern.ch>