Onboard basic RAG preset

[ohltyler]

Description

Adds a basic RAG preset to the library. Pre-populates:

• ML search response processor with an LLM to generate results (defaults to many-to-one)
• collapse response processor to only return unique ML responses (in many-to-one case, this will be one result)

More details:

• onboards collapse search response processor
• adds an llmResponse quick-config field, and default quick-config field values for the new RAG usecase
• fixes bug of input transform fetching failing on search response processors when no prior search request or search response processors
• refactors helper fns in QuickConfigureModal and makes them more generic

Notes:

• future PRs will improve the prompt template editing experience. For now, it is entirely free-form.
• future PRs will improve the checking of full_response_path when implementing transforms with JSON path. See tracking issue Enforce full_response_path=true when users configure JSON Path in ML processors #368
• future PRs will improve the many_to_one/one_to_one toggling experience. See tracking issue Support many-to-one in output transform modals for ML search response processors #367

Demo video, showing a RAG usecase using a deployed LLM model with some sample review documents. The model does not have an interface or configured prompt template, so both are manually filled in. The final result is a single response containing the LLM's summary of the documents found during search.

screen-capture.17.webm

Check List

• Commits are signed per the DCO using --signoff

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[ohltyler]

This is the last "building block" use case. After this is merged, we can explore common combinations - semantic search with RAG, hybrid search with semantic search and semantic analysis, hybrid search with plaintext search and semantic analysis, with RAG, etc.....