feat(gemini): image reasoning demo with ai sdk v7 beta

summary

• upgrade ai sdk from v3 to v7 beta with streamText, useChat, and reasoning-file support
• add gemini 3.1 flash image and gemini 3 pro image models via ai gateway
• real-time streaming of reasoning text and draft images
• 3-draft image exploration in reasoning before final selection
• multi-turn image editing support
• next.js 16 with proxy instead of middleware
• slash commands (/new, /clear, /delete, /theme, /model, /help)
• keyboard shortcuts (Cmd+N, Cmd+K, Cmd+Shift+Backspace)
• image lightbox with download
• copy button on ai responses
• chat rename via pencil icon in sidebar
• botid bot protection on /api/chat
• vercel analytics
• ai-generated chat titles with gemini 2.5 flash
• smart routing: text-only responses for non-image prompts
• loading states, error boundaries, blocked prompt detection
• removed auth, flight booking tools, file upload