[FEATURE] Make it possible to generate images

[edenreich]

Summary

Some models like Nano Banana from google can generate images. The goal here is to allow the model to generate an image based on a prompt.

Acceptance Criteria

• The CLI can generate images using models like Nano Banana
• A follow up ticket is created on a related repo (if applicable - perhaps something on the gateway might need to be adjusted)
• It's tested
• It's documented

[edenreich]

I think the current best approach for this would be to create an A2A agent that generates images from from a prompt.

The reason is because this functionality only supported by some providers (OpenAI or Google).