Rid vs. Id consistency needs to be enforced

### Bug Title

Rid vs. Id consistency needs to be enforced

### Crate Name

azure_data_cosmos_driver

### Crate Version

n/a

### Description

ntended-rid correctness is incomplete.
The key header is x-ms-cosmos-intended-collection-rid. The driver currently lacks a dedicated stale-container handling path keyed to mismatch substatus flows (Python recreate flow is keyed around
COLLECTION_RID_MISMATCH (1024) / THROUGHPUT_OFFER_NOT_FOUND (10004) and related stale-cache signals) for a stable error contract. The driver should also stamp intended-rid autonomously for container-scoped requests it builds.
Impact: rare but highest severity (wrong-container risk in drop+recreate workflows).

### Steps to Reproduce

n/a

### Checklist

- [x] Follow our [Code of Conduct](https://github.com/Azure/azure-sdk-for-rust/blob/main/CODE_OF_CONDUCT.md)
- [x] Check that there isn't already an issue that request the same bug to avoid creating a duplicate.
- [x] The provided reproduction is a [minimal reproducible example](https://stackoverflow.com/help/minimal-reproducible-example) of the bug.

Start in the azure_data_cosmos_driver crate by tracing container-scoped request construction and the existing stale-container or mismatch error handling. Compare the intended-rid header behavior with the documented COLLECTION_RID_MISMATCH (1024) and THROUGHPUT_OFFER_NOT_FOUND (10004) flows; done means requests stamp the intended RID and mismatches have a stable error path.

Written by the indexing model from the issue text.