[Feature Request]: Metadata Cleanup

[strawgate]

Feature Description

When using docling I'm ending up with 25kb documents as docling pushes a lot of junk into metadata and the relatednodeinfo object brings related node metadata into the node meaning there are 3-4 copies of this metadata

I have a metadata cleanup transformer which I'm using to clean this up

Is something worth contributing?

Reason

No response

Value of Feature

Much faster search and ingest as document sizes are much smaller

[logan-markewich]

Probably makes more sense to update the docling integration to have more configurable metadata? Or at the very least remove the stuff thats definitely junk by default