- Add support to retrieve the color model and bit depth of the original file (before conversion to RGB)
- allow to load only metadata fields and not the actual image data
- Add a method to retrieve a unique hash from a file. This hash is very much like what Amarok uses for file identification, and the purpose is to be fast and unique.

Similarity detection is _not_ possible with this hash - for this, we will use a Haar-like algorithm.