Set up a Scholia-specific SPARQL endpoint

[Daniel-Mietchen]

The main motivation behind this is that this would allow us to run queries that time out on the official endpoint.

Of course, this would be more resource-intensive, so such queries

• would likely not (normally) be run in a user-triggered fashion
• may not (normally) be synced to the live Wikidata (but work on dumps instead, or perhaps on LDF)
• would be cached (cf. Cache? #67 ).