To mount the filesystem run
  $ ./mount.bzrfs <path-to-branch> <mount-point> -o log=/tmp/bzrfs
(the log-option writes debug-output to the given file. there is
currently a lot of debug-output, i will reduce this and add a better
logging structure later)

To unmount run:
  $ fusermount -u <mount-point>

The filesystem's structure is as followed:
  * There are three main folders:
      ** REVISIONS: this folder is completely read-only and contains
         the different revisions of the branch. With
            '-o extract_revisions=[start:]stop[:step]'
         you can reduce the number of shown revisions at mount-time
      ** CURRENT: this contains the current workingtree. You have
         read/write access to this directory
      ** TAGS: Same as REVISIONS but with the tags added to the branch
         as subdirectorys. This is currently not implemented

At the current state you should be able to add/change files and
directories in the CURRENT folder.

Future ideas:
   * add a .STATS directory to each revision in REVISIONS, this folder
     contains files like .DIFF (diff against the previous revision) and
     .COMMENT (commit message for this revision)
   * add a 'bzraction' commandline option to the mount script, which
     can have the following values:
      ** IMMEDIATELY: whenever the content of CURRENT changes the
         related bzr action will be performed (e.g. when you add a file,
         a command like 'bzr add' will be executed
      ** MAGIC: perform bzr-commands via magic-virtual-files
         (e.g. '$ echo boo bar baz > .commit' will commit the latest
         changes with message 'boo bar baz')
      ** TOOLS: you can run scripts within the filesystem to perform
         the bzr actions (e.g there will be a script 'bzr-add' which
         wraps the 'bzr add' command for the bzrfs filesystem)