add gemini image models in .chat

[P3il4]

added support for gemini models in .chat that can edit images in a conversation
similar to how openrouter does it (image_config, response formats, etc)
implementation based on gemini image provider
kept the same usage keys as txt2img (gemini:model:output:text/image or input)
testing:

await puter.ai.chat([
    { role: 'user', content: 'Draw a cat' },
], { 
  model: "gemini-3.1-flash-image-preview", 
  stream: false, 
  image_config: { aspect_ratio: '16:9', image_size: '1K' } 
});

for stream: false, image is in result.message.images
for stream: true, in image type event
also works with openai compatible route, matching openrouter format exactly

const apiBase = puter.APIOrigin;
const token = puter.authToken;
const resp = await fetch(`${apiBase}/puterai/openai/v1/chat/completions`, {
    method: 'POST',
    headers: {
        'Content-Type': 'application/json',
        'Authorization': `Bearer ${token}`,
    },
    body: JSON.stringify({
        model: "gemini-3.1-flash-image-preview",
        messages: [{ role: 'user', content: 'Draw a small red dot' }],
        stream: true,
        image_config: { aspect_ratio: '16:9', image_size: '1K' }
    }),
});

there are more examples in docs but those are the main ones
tested with a mock server, would appreciate testing on a real key.

[ProgrammerIn-wonderland]

This is awesome

[ProgrammerIn-wonderland]

I see thoughtsignature is fixed now, testing again