### Check for existing issues

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

### What happened?

For OpenRouter-routed image models (e.g. `openrouter/google/gemini-3-pro-image-preview`, `openrouter/google/gemini-2.5-flash-image`), the `image_config` parameter (`aspect_ratio`, `image_size`) is **honored on image generation** but **silently ignored on image edits**. The edit returns an image at the *source* image's dimensions regardless of the requested `image_config` — no error, just a no-op.

Generation and editing both go through the same OpenRouter chat-completions bridge, and OpenRouter accepts `image_config` for both — but LiteLLM's edit path strips it before the outbound request is built.

## Expected

`litellm.image_edit(model="openrouter/...", image=..., prompt=..., image_config={"aspect_ratio": "16:9", "image_size": "2K"})` should forward `image_config` to OpenRouter, exactly as `litellm.image_generation(...)` does — producing an edited image at the requested aspect/size.

### Steps to Reproduce

```python import litellm

# Generation — image_config IS honored (output is 16:9 / 2K) litellm.image_generation( …