* Rework metadata structure in processing messages to be a subgraph
* Add subgraph creation for tg-load-pdf and tg-load-text based on command-line passing of doc attributes
* Document metadata is added to knowledge graph with subjectOf linkage to extracted entities
