[META] ML Inference Processor Enhancements I

[mingshl]

Is your feature request related to a problem?

Search response processor:

• support one to one inference in ml inference search response processor
• support list in substring during prediction API, so it can support GenAI/RAG use case in ml inference search response with prompt defined in connector level
• support custom prompt in model config, currently it will cause escape problem, due to model Input change in this issue: [BUG] model_config prefix is changed in payload for ML inference processors  #2822

What solution would you like?
A clear and concise description of what you want to happen.

What alternatives have you considered?
A clear and concise description of any alternative solutions or features you've considered.

Do you have any additional context?
Add any other context or screenshots about the feature request here.

[mingshl]

@opensearch-project/admin -- Can you please move this to our roadmap https://github.com/orgs/opensearch-project/projects/1 under 2.17?

[mingshl]

support one to one inference in ml inference search response processor in #2801

support list in substring during prediction API, so it can support GenAI/RAG use case in ml inference search response with prompt defined in connector level in #2871

adding a to_string() method in httpConnector to support custom prompt in #2871