Speed up tripper.dataset.save_data()

[jesper-friis]

The tripper.dataset.save_data() creates a json-ld string that is sent to rdflib.Graph.parse(format="json-ld") for each documented resource. This is very slow for when documenting many resources.

This can probably be speeded up a lot if we documents all the resources in a single json-ld and only calls rdflib.Graph.parse(format="json-ld") once. The json-ld document could use the "@graph" keyword like this:

{
  "@context": "...",
  "@graph: {
    "iri_for_dataset1": {...},
    "iri_for_dataset2": {...},
    ...
  }
}