Be more specific with mime types in accept headers to sparql endpoints

[jaw111]

Related to #304 some of our users have begun to use the .csv extension on the request URLs to directly access the results in CSV format.

We are using Amazon Neptune and today observe that these requests suddenly start returning JSON results after upgrading the cluster to the new version 1.0.4.2 engine. On closer inspection it transpires that grlc makes those requests with header Accept: text/csv; q=1.0, */*; q=0.1 and the new Neptune version gives a JSON response for those.

To help avoid such nasty surprises, it would appear wise to be more specific in the accept headers that are sent in sparql protocol requests. My suggestion is to modify mimetypes as follows to be much more specific in the media types and extend support for additional (standard) formats:

Format	Accept
csv	text/csv; charset=utf-8
tsv	text/tab-separated-values; charset=utf-8
json	application/sparql-results+json
srj	application/sparql-results+json
xml	application/sparql-results+xml; charset=utf-8
srx	application/sparql-results+xml; charset=utf-8
ttl	text/turtle
nt	application/n-triples
rdf	application/rdf+xml; charset=utf-8
jsonld	application/ld+json
html	text/html

Note the use of charset parameter on those media types where the character set is not implied by the media type.