Multi modal RAG benchmark

[rlancemartin]

• Example notebooks for eval of multi-modal RAG w/ mm-embd and mv-retriever vs baseline top-k RAG
• To Do: create public benchmark

[hinthornw]

Ideally we don't check these into the repo and either use LFS or GCS

[rlancemartin]

Ya, I'm removing now that we are creating the public dataset.

[hinthornw]

Fixed up the paths and notebook

[rlancemartin]

Fixed up the paths and notebook

Nice. I will further clean both both notebook and add comments / context.