Implement a MultiDataStore

The idea of a multidatastore is to decorate two datastores. It can be used to implement a Primary/Archive DataStore.
New binary files are saved into the primary. A archive task would move files from the primary to the backup datastore. Reading from a MultiDataStore is transparent to the jcr api. So it will always return files either from the primary or from the backup datastore.
The primary datastore can be mapped to a fast storage medium and the archive to a cheap sata or a tape library.