Allow documents to be loaded locally for Knowledge Injection

[murthyrudra]

This feature allows users to specify the local folder path instead of github repo for knowledge injection. Specifically, the update modifies the validate_document function to search for files locally and ensure the documents follow the prescribed format.

This feature includes an update to the schema repo and the corresponding PR is here

[bbrowning]

The change to the schema was never approved/merged, so I'm holding off reviewing the implementation here until it's decided to support local knowledge docs at the schema level.

[bbrowning]

Given that we're making a plan right now to move most of this document ingestion code to the instructlab/instructlab repository, and that there's no movement on the pre-requisite PR in the schema repo, I'm closing this for now. If or when a new knowledge schema gets created that supports local knowledge documents, we're happy to accept contributions or work together on the implementation of that. Thanks!