### Check for existing issues

- [X] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

When creating an S3-backed vector store from the dashboard, the selected embedding model is forwarded to `ragIngestCall` under `ingest_options.vector_store.embedding_model`. The RAG ingestion code only reads embedding settings from top-level `ingest_options.embedding`, so it never sees the selection and defaults to `text-embedding-3-small`. The endpoint persists the user-selected model for later search, so the vector store is ingested with one embedding model and queried with another.

**Observed:** `S3VectorsConfig` writes the selected model into `providerParams.embedding_model`. `ragIngestCall` places `providerParams` under `ingest_options.vector_store`, but `BaseIngestion` reads embedding configuration from `ingest_options.embedding` only. `S3VectorIngestion` therefore falls back to `{"model": "text-embedding-3-small"}` for ingestion. The RAG endpoint stores the user-selected `embedding_model` in `litellm_params`, and `S3VectorsTransformation` uses that value (the user-selected model) at query time. Ingested document vectors and query vector…