Text to Image: Support image generation as an AITool

[jeffhandley]

[ericstj]

Open AI's reponses client natively supports an ImageGen tool, but I've had trouble making it work well. Here's my prototyping branches:
https://github.com/ericstj/extensions/tree/ImageGenTool
Sample

I need to root cause and follow up with OpenAI or Azure.OpenAI about the issues I hit.

There are also other chat models that have similar support - for instance https://ai.google.dev/gemini-api/docs/image-generation. While not a "tool" per-se, we could expose the functionality in this way for consistency.