Implement graph aggregation node/relationship discovery callback

[jdcasey]

This would allow parallel processing of discovered nodes / relationships while discovery is still in progress. Traversing / processing the graph after discovery seems to take about half the time of a total operation like "downlog" or "urlmap" (these are currently AProx wrappers around ResolveOps.resolveRepositoryContents(), see #35 ).

The graph-aggregation process still iterates existing nodes / relationships when resuming discovery with an existing graph. We'd have to have some special-case handling logic to accommodate this part, in addition to callback triggers for newly discovered nodes/relationships.