[Feature] Image generation tool

[Rustam-Z]

Problem

The bot can render HTML→PNG and LaTeX→PNG but has no way to generate images from text prompts. Users asking for "generate an image of X" get nothing.

Proposed solution

Add a generate_image(prompt, style?) MCP tool:

• Backend options: DALL-E 3 (OpenAI), Imagen (Google), Stable Diffusion (self-hosted)
• Returns a file path → passed to send_photo for delivery
• Config: IMAGE_BACKEND=dalle|imagen|sdxl, IMAGE_API_KEY

Notes

• Keep render_html / render_latex for charts/math — they're more reliable for structured visual output
• Image generation is for creative/illustrative requests
• Operator can disable via plugins.json flag