Support updateGraph concurrency

In the present implementation, with concurrent calls to updateGraph (i.e., concurrent link-indexer runs), multiple base nodes (`Node`) having the same identifier (URL) may occur in the graph database.

Start by locating updateGraph and the Node model or persistence entry point; inspect how concurrent calls create base nodes identified by URL. Done means concurrent link-indexer runs no longer leave multiple base nodes with the same identifier in the graph database.

Written by the indexing model from the issue text.