Distinguish machine-readable and web representation of distribution

[ddeboer]

There’s some confusion as how to apply schema:contentUrl to distributions. For example the NA distributions have SPARQL access URL https://www.nationaalarchief.nl/onderzoeken/sparql, but that’s the UI. The actual machine-usable SPARQL endpoint is https://service.archief.nl/sparql (netwerk-digitaal-erfgoed/dataset-register#352).

The same may apply to data dumps:

Proposal: schema:contentUrl must be the machine-readable URL. Change the requirements:

- The URL where the distribution can be directly accessed.
+ The URL where the distribution can be directly accessed by machines. 
+ For a URL pointing to a web page describing the download or a (SPARQL) user interface, use recommended property `schema:???`.

@coret Should we use schema:mainEntityOfPage, schema:url or some other property?