The conditions used to determine whether or not an image have to be
processed depending on its size were boggus. This commit fixes the
issue, images are now processed as they should.
Before this commit, every images were reprocessed all of the time.
Because the address book may be re-compressed when new pictures are
added, we want to process them and not reprocess the previous ones.
This commit include a size detection so images below the size limit
are not processed.
