Almost there

This patch does the following things:
- The _NO_STAT flag has been added.
- The _FOLLOW_SYMLINKS flag has been added in the follow-symlinks branch and
  is not intended for this development cycle
- Remove TrackerDataProvider's _{get|set}_dir_flags() API (no longer used).

This fixes a warning about not catching a potential GLib.Error being thrown
for the GUnixFDList's _append() API.

Now we only show common RDF types by default, e.g. nfo:Document and it is
based on .rules file fallback types and some others specifically added.

There is also now a --all option to return all stats as was the behaviour
before.

In addition to this, the EXPRESSIONS provided on the command line following
the OPTIONS can be used to filter the stats, e.g. : 'tracker-stats doC' will
list the nfo:Document class and resources in the DB matching.

GMarkup is really not that good at parsing XML, so we need to try
harder to ignore errors parsing the contents of EPub files, and
populate the index with *some* data.

https://bugzilla.gnome.org/show_bug.cgi?id=735645

Error extracting EPUB contents (OEBPS/Text/info.xhtml): Error on line 59: Entity name 'copy' is not known
is better than:
Error extracting EPUB contents: Error on line 59: Entity name 'copy' is not known

https://bugzilla.gnome.org/show_bug.cgi?id=735645

All of the XDG user-dirs will be set to the content directory's
path, so that applications such as gnome-documents can find the files
in the expected directories.

https://bugzilla.gnome.org/show_bug.cgi?id=735636

This was causing Valgrind to report reading past memory barriers

Should be nfo:EBook now, not nfo:TextDocument.

The --disable-miner option doesn't exist anymore.

For which metadata extraction isn't currently available.

https://bugzilla.gnome.org/show_bug.cgi?id=735460

Through the dummy extractor

https://bugzilla.gnome.org/show_bug.cgi?id=735460

https://bugzilla.gnome.org/show_bug.cgi?id=735460

For use with data types that don't have any additional metadata
inside the file, but need tagging with specific RDF types.

Note that the use of a dummy extractor is only temporary, and rule files
should instead allow not having a ModulePath entry.
See https://bugzilla.gnome.org/show_bug.cgi?id=735610

https://bugzilla.gnome.org/show_bug.cgi?id=735460

Martyn Russell authored Aug 28, 2014 and Bastien Nocera committed Aug 28, 2014

https://bugzilla.gnome.org/show_bug.cgi?id=735460

https://bugzilla.gnome.org/show_bug.cgi?id=735567

Previously we would not call this API and it meant that any implementation
with memory allocated waiting to be cleared up, would not have been freed. Now
we call this API just about the same time we signal up the stack that we've
::directory-crawled

The leak occurred because tracker_sparql_task_new_with_sparql() was being
called but the returned TrackerTask* was not being unreferenced anywhere and
the call to tracker_sparql_buffer_flush() with the new task was taking its own
references internally.

Took this opportunity to make the code here easier to follow:
- do_process_file() is now merged into item_add_or_update()
- item_add_or_update_cb() is renamed to item_add_or_update_continue() so it's
  obvious it is called from tracker_miner_fs_file_notify().
- renamed various variables to make the code easier to follow.