New Feature: Run KG extraction on already imported structured data (w/text)

[jexp]

• If someone has already text data in a number of properties of nodes in neo4j (e.g. imported from external structured data), can we enrich or mark up that data to be processed by the KG builder?
• In principle we need a "virtual" document and an additional "__Chunk__" label on the nodes with the texts. and then use the "pick up from where it stopped"
• usually those text parts are smaller than our chunk size so we might be able to skip chunking itself
• so we only need chunk embeddings, entity extraction as the minimal steps