### Check for existing issues

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

### What happened?

When using stream=True with Scaleway (or any provider returning a reasoning field within the delta object), LiteLLM crashes with a TypeError.

The issue occurs because the streaming parser/proxy fails to gracefully handle the non-standard reasoning key inside the delta object during an asynchronous iteration, resulting in the response object becoming None.

### Steps to Reproduce

1. Use LiteLLM to call a Scaleway model with stream=True. 2. The provider returns chunks where the delta contains a reasoning key (e.g., {"delta": {"reasoning": "..."}}). 3. The async iterator fails.

```py import litellm import asyncio

async def test_stream(): # This call triggers the error response = await litellm.acompletion( model="scaleway/gemma-4-26b-a4b-it", messages=[{"role": "user", "content": ""}], stream=True ) async for chunk in response: print(chunk)

asyncio.run(test_stream()) ```

**Expected** LiteLLM should either: Support the reasoning field in the delta object as part of the stream. Or, gracefully cat…