Fix the logic behind matching graph identifiers with the ones coming from the aggregation subsystem

[marekhorst]

In the #1361 pull request we have replaced the ObjectStore as the source of contents metadata with the hive-based aggregation subsystem.

The identifier matching strategy was simplified, as described in this comment but it turned out we should still involve originalId field when matching the aggregation subsystem entries with the graph results.

This means we should reintroduce the originalId field in the joining operation in order to increase the overlap between the aggregation subsystem payload identifiers and graph result ids.