Consider switching to RDF4j background parsing implementations

[RickMoynihan]

We currently consume results on a thread and use a pipe to convert them into a lazy sequence of results.

Since we've done this RDF4j has added convenience methods for pull based parsing (that do the same thing underneath). We should assess whether we can use them, and consider replacing this:

grafter/src/grafter/rdf/io.clj

Lines 535 to 562 in 66d80c3

	(pr/to-statements [reader {:keys [format buffer-size base-uri] :or {buffer-size 32
	base-uri "http://example.org/base-uri"} :as options}]
	(if-not format
	(throw (ex-info (str "The RDF format was neither specified nor inferable from this object.") {:error :no-format-supplied}))
	(let [[statements put!] (pipe buffer-size)
	parser (doto (fmt/format->parser (fmt/->rdf-format format))
	(.setRDFHandler (reify RDFHandler
	(startRDF [this])
	(endRDF [this]
	(put!)
	(.close reader))
	(handleStatement [this statement]
	(put! statement))
	(handleComment [this comment])
	(handleNamespace [this prefix-str uri-str]))))]
	(future
	(try
	(.parse parser reader (str base-uri))
	(catch Exception ex
	(put! ex))))
	(let [read-rdf (fn read-rdf [msg]
	(if (instance? Throwable msg)
	;; if the other thread puts an Exception on
	;; the pipe, raise it here.
	(throw (ex-info "Reading triples aborted."
	{:error :reading-aborted} msg))
	(sesame-statement->IStatement msg)))]
	(map read-rdf statements))))))

with something using this