hash-based blob ids

[retog]

Rather than storing blobs with some random identifiers Tridoc-Backend should compute a hash slash digest based on the content of the file. It should still be possible to have the same file with different titles and different tags. For that a separation between logical document and blob should be introduced in the metadata.

Example:

<http://3doc/data/PH5aeqVDU2A54GULooh~8>
        a                               <http://schema.org/DigitalDocument> ;
        tridoc:blob "QmNrgEMcUygbKzZeZgYFosdd27VE9KnWbyUD73bKZJ3bGi" ;
[...]

To compute the hash the algorithm used by IPFS shall be used, implemented e.g. here: https://github.com/alanshaw/ipfs-only-hash.

A rest endpoint /migrate shall be introduced to migrate existing instances.