Expose RAG_USE_FULL_CONTEXT in UI

[ala85]

@danny-avila I d like to help exposing that as UI parameter, but before jumping into code I d like to ask:

1. Where would you recommend to have the compoent placed in UI (Probably similar to "Resend Files")?
2. The chat endpoint requires this information and I was leaning towards adding it as its own value (not as part of TMessage) do you agree?
3. Any other things to be aware?

Thanks

[danny-avila]

The better thing to work on is the following:
#7135

Also an OSS alternative to mistral OCR would be preferable to work on, which mistral OCR integration sets up the scaffolding for.
This one seems like the best to support, but could be any whose responses could be transformed to match the expected output for LC:
https://github.com/docling-project/docling-serve

The reason is that feature was experimental from the RAG API. Also, it's not guaranteed that the "full context" would be in order without some changes to the RAG API itself. That could also be another problem to tackle.

[ala85]

I ll look into that, but may still be worth to pursue when installation of custom OCR may not be immediately possible (e.g enterprise, incompatible OCT, etc..) and seems a low hanging fruit, given rag file has already a naive implementation which still seems to work given my tests.