Dedicated RdfParser for NTriples

[aindlq]

Just a random thought.

I think it would be great to have a dedicated RdfParser for NTriples.

I think it can be a bit confusing to conflate turtle and ntriples when building the index. See #1544

Actually for proper ntriples one can have parallel processing without regex checks and also doesn't need to care about prefixes in the parser, so the whole parsing can be even more faster. Typically big datasets provide data in ntriples, because it is the most efficient for parsing.