What is the intended role of the various URLs for a dataset?

[ldodds]

The code has various URIs for a dataset:

• access_url -- which is the url given by the user, however the data might ultimately be loaded from a different location
• uri/url -- I think these are always aliases of the above, so could be removed?
• dataset_uri -- the actual URI of the dataset, which may/may not be the same as the above

It might be simpler to go with a distinction of:

• access_url -- location from which metadata is loaded; this might get updated to reflect, e.g. autodiscovery links
• uri -- the identifier for the dataset, derived from the graph or other metdata

What do you think?