New Feature: handling schema changes for an already extracted graph

[jexp]

• as we still have all the chunk data, we can re-run extraction and post-processing
• question is what happens with the existing entities do we need to keep them or remove them or merge the two processing steps together?
• depends probably also on the schema change, if it's additive or subtractive
• should we store the used schema on the document node? like we do for model and other parameters?

[bengentan]

This is a very useful scenario, as the source might be changing.
If we built the nodes/relationships with LLM, I am just curious: how do we find the right node/relation we need to update? LLM could generate different entities with the same context.