Add separate metadata for each endpoint

[acharal]

Currently the metadata of all endpoints in the federation are in a single turtle file expressed in SeVod (an extension of VoID). It would be useful to provide the flexibility to separate the access of metadata per endpoint.

This separation will give the ability for combining mixtures where some sources serve their own metadata. Another use-case is a scheduled job that scrapes a source and updates their summaries in another (remote but accessible) store.