test(openai): point image extra_headers coverage at with_raw_response

Main now reads provider headers off the raw response, so the proxy
credential cases mock that call and pass a real ImageResponse.

Co-authored-by: Zsanz3 <Zsanz3@users.noreply.github.com>

tests/unit/llms/openai/image_generation/test_openai_image_generation_extra_headers.py

@@ -275,7 +275,7 @@ class TestImageGenerationExtraHeaders:
             await openai_chat_completions.aimage_generation(
                 prompt="A white cat",
                 data={"model": "gpt-image-2", "prompt": "A white cat", **optional_params},
-                model_response=MagicMock(),
+                model_response=ImageResponse(),
                 timeout=60.0,
                 logging_obj=mock_logging_obj,
                 api_key="test-key",