### Check for existing issues

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

### The Feature

### Summary

GenericGuardrailAPIRequest (the wire payload posted to a generic guardrail provider's apply_guardrail endpoint) doesn't carry the original chat-completion request's stream: true/false flag. This makes it impossible for a guardrail provider to tell at request time whether the LLM call it's running on is streaming. For some guardrail use cases that distinction matters.

### Proposed change

Add a single boolean field to the wire payload:

class GenericGuardrailAPIRequest(BaseModel): ... is_streaming: bool = False

Populate it in apply_guardrail() (in litellm/proxy/guardrails/guardrail_hooks/generic_guardrail_api/generic_guardrail_api.py) by reading request_body.get("stream", False). Construction site is around the existing GenericGuardrailAPIRequest(...) call (~line 436 on current main).

### Backwards compatibility

- Receiver side: Pydantic models on the receiving service ignore unknown fields by default, so older guardrail providers running against newer LiteLLM versions are unaffected. - Sender side: Default value False matc…