question as to use cases for the "huddle" use of bnode identifiers

I would find it very useful to have some fully-worked-out use cases for the
"huddle" treatment of bnode identifiers in SPARQL query results.  These use
cases should also address virtual graphs, i.e., graphs that are the result of
computations (including various entailment-preserving operations like leaning)
and that thus do not correspond directly to input graphs.  

peter