Adding Neural Sparse Search preset

[saimedhi]

Description

• Onboarding a new preset for Neural Sparse Search.
• For development used "opensearch-neural-sparse-encoding-v2-distill" deployed on sagemaker endpoint.
• Query used in Search Flow is neural_sparse query. (Search the data using a sparse vector)
• I will upload video.

Issues Resolved

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]

First pass, overall looking good! Main concern I have is on the usability, and the fact we can't point to easily deployable neural sparse models. We need to ensure that added presets are up to a high bar, such that choosing them will likely mean minimal user inputs to get started.

The fact that there is no standard model (or pattern of inputs/outputs) makes it difficult to automatically configure the input/output mappings of the processors as well.

Let's discuss if there should be added callouts/wording around this particular use case, and have good documentation we can point to regarding the model setup (thanks for adding the proposed models and tutorial!)

[brianf-aws]

Currently this workflow will be blocked before we merge this Adds Json Parsing to nested object during update Query step in ML Inference Request processor #3856 The limitation was it could not parse the sparse vector to perform neural sparse search.