Unique identification of datasets across data points

[c-martinez]

With the current specification of the FDP, it looks like one data set which is indexed on two different FDP's could end up having different identifiers (and thus could incorrectly be assumed to be two separate data sets).

E.g: I could deposit my-dataset on FDP1 as:
http://fdp1.org/fdp/dataset/my-dataset-X

and on FDP2 as:
http://fdp2.org/fdp/dataset/my-dataset-Y

This should perhaps be adjusted on the specification (not necessarily on this repo).