query :: conversion to dataframe is really slow

[mpo-vliz]

Trying out the sema.query (previous pykg2tbl with some larger resultsets making dumping (via pandas.Dataframe) to csv run for ever.

Detail logging shows the time seems to be spent in the conversion from query-result into dataframe.

We should look into making that more efficient (probably by doing less in memory copying?)
Apparently there has been some work in this are:

• convert query results into pandas dataframe RDFLib/sparqlwrapper#125 (comment)
• Convert rdflib.plugins.sparql.processor.SPARQLResult to a pandas.DataFrame? RDFLib/rdflib#1179 (comment)

[mpo-vliz]

applied in 99263fd

the positive effects on performance are nothing less than dramatically enormous --> for a large resultset (150k rows) this cut the conversion to df from >800k millis to a mere 4 millis!