[META] Generic ML inference processor for ingestion pipeline

[ylwu-amzn]

Today we have specific ML/AI ingest processor for different use case. For example:

1. TextEmbeddingProcessor for generating embeddings with text-embedding models
2. GenerativeQAResponseProcessor for QA with LLM
3. PersonalizeRankingResponseProcessor for reranking search result with AWS personalized Service model.

Users have more and more use cases like language identification, NER etc. To avoid the effort to build more and more specific processors, we can build a generic ML inference processor which can enrich data with any ML model.

[ylwu-amzn]

PR merged #2205