cbtagfs limitations
-------------------

 - No support for complex tag queries.

   You can filter by as many tags as you want, but they are all joined by
   logical "AND"s.

 - No support for detecting source file renames or moves.

   cbtagfs works best on unchanging data. If you start moving source files
   around, the symbolic links that cbtagfs serves will be broken.

   I would love to add better support for intelligently detecting renames/moves
   in the future by a combination of inode and filename comparisons, but the
   support is lacking now.

 - All indexed filenames must be globally unique.

   This is due for fixing soon. A directory can't serve two files with the same
   name, and cbtagfs doesn't have any logic for detecting duplicates and making
   them appear unique. Future versions of cbtagfs will probably append a
   sequence number to twin filenames within results.