Support for in-memory tier feature store with offline feature store

### Product Version

- [ ] Amazon SageMaker Studio Classic
- [ ] Amazon SageMaker Studio
- [x] It is not related to SageMaker Studio

### Product Category

Feature Store

### Description

A key advantage of a feature store is to avoid online / offline skew. However, Since there is [no buffering possible](https://docs.aws.amazon.com/sagemaker/latest/dg/feature-store-storage-configurations-online-store.html) from the online in-memory store to the offline feature store, that advantage is gone. I don't want to build custom logic to move data from online to offline store.

You can also find the problem mentioned [here](https://www.reddit.com/r/mlops/comments/1p8qnm4/the_drawbacks_of_using_aws_sagemaker_feature_store/), along with scaling problems. Seems like a lot of people wish for an improved feature store, also with regard to more scalable ingestion.

### Other Details

_No response_

No repository files, tests, or implementation entry points are identified. Start by reviewing the linked SageMaker online-store documentation and the referenced discussion to clarify buffering and offline-store requirements. Done means the requested in-memory-to-offline feature-store capability has agreed scope and acceptance criteria.

Written by the indexing model from the issue text.