OWL 2 RL materialization with [reasonable], for native backends.
materialize reads every asserted triple (all graphs, merged), runs the reasonable
Datalog engine in memory and writes the new conclusions into quads_inf (graph 0), exactly
where the SQL rules of oxilite_core::reason put theirs. It is much faster than the SQL
fixpoint on large data but needs the whole dataset in memory, so it is not used on D1.