Previously TagExtractor.parse_text would return all matches, including
duplicates, and would therefore potentially link a tag to a set or
sample multiple times (if used mutiple times in the info field).

This commit ensures tags are extracted only once, but will not affect
the link in the frontend

tags can now be queried and auto-completed by multiple group_ids. This
is in order to better manage tag auto-completion for samples added to multiple groups as well as auto-completion for search bars

Tag searching is now managed in the search box entirely. The contents is
filtered and separated into tags and search strings in order to provide
an easier way for users to manipulate tags and filter results in a
seamless manner.

the object is to filter the search bar and and extract tags from
searches

refactores the url sanitisation to the TagDecorator, since we had a
little code duplication

renaming decorate_text to decorate slighlty reduces the length of lines
while remaining sufficiently clear as to the purpose of the function, I
believe.

This is to give access to web2py HTML helpers for some refactoring

Although the jquery plugin autocomplete.js was providing the basic
functionnality we needed, it was not natively able to display the list
of suggestions in the desired position for our textareas.

At.js is capable through the use of Caret.js of displaying a suggestion
box under the caret in a textarea.

This javascript code is heavily inspired by the code present in the
gitlab-ce project and allows us to have a nicely cached data-set to
limit database queries.

register the tags used by a group when registering tags to records

creates a link that actually functions within the vidjil app by creating
a callback which is triggered when clicking on a tag url.

This module will allow the automatic creation and decoration of tags
within selected text fields.
A database entry is created for each new tag which will then be linked
to database records. This will later allow for easy querying.