Add AI-powered semantic SVG recolorization

[Copilot]

Current distance-based (CIEDE2000) color matching doesn't understand semantic roles—a logo's primary brand element and decorative background might map to the same palette if their colors are similar.

Changes

New AI mode (recolorizeSvgWithAI):

• Analyzes SVG via AI vision APIs (OpenAI GPT-4o, Anthropic Claude, or custom)
• Maps elements to semantic roles (primary-object, background, accent, etc.)
• Assigns palettes based on conceptual meaning, not color distance
• Falls back to distance-based on error

Enhanced recolorizeSvg:

• Optional aiConfig parameter enables AI mode
• Returns Promise<string> when using AI, string otherwise
• Zero breaking changes—existing code unchanged

Providers supported:

• OpenAI (gpt-4o)
• Anthropic (claude-3-5-sonnet)
• Custom endpoints

Example

// Before: distance-only (still default)
const result = recolorizeSvg(svg, palettes);

// Now: semantic understanding via AI
const result = await recolorizeSvg(svg, palettes, {
  aiConfig: {
    provider: "openai",
    apiKey: process.env.OPENAI_API_KEY,
  }
});

AI identifies elements by role, e.g.:

[
  {"selector": "circle[cx='50']", "role": "primary-object"},
  {"selector": "rect[width='200']", "role": "background"}
]

Maps to palettes: primary-object → primary, background → surface, accent → tertiary.

Implementation notes

• Brightness-based tone selection within chosen palette
• Automatic fallback prevents service disruption
• 16 tests covering all providers and error paths
• Docs: docs/AI_RECOLORIZATION.md, docs/MIGRATION_GUIDE.md

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.