feat: add OpenAI-compatible provider support

[hongkongkiwi]

Summary

Adds a configurable AI provider switch so the addon can use either:

• Gemini-compatible models (existing behavior)
• OpenAI-compatible endpoints (OpenAI / OpenRouter / Z.ai / etc.)

This lets users choose between provider ecosystems without changing addon logic.

What changed

• Added provider abstraction + OpenAI-compatible Chat Completions adapter (utils/aiProvider.js)
• Updated addon AI calls to use the selected provider (addon.js)
• Updated /aisearch/validate to validate the selected provider (server.js)
• Updated configuration UI to select provider + show relevant fields (public/configure.html)
• Updated docs/landing copy to reflect provider choice (README.md, index.html)

Configuration

Gemini

• Provider: gemini
• Key: Google AI Studio key
• Model: e.g. gemini-2.5-flash-lite

OpenAI-compatible

• Provider: openai-compat
• Key: provider API key
• Base URL examples:
  • https://api.openai.com
  • https://openrouter.ai/api (or https://openrouter.ai/api/v1)
• Model examples:
  • gpt-4o-mini
  • openai/gpt-4o-mini (OpenRouter-style)

Backwards compatibility

• Existing configs without AiProvider continue to default to Gemini.

Test plan

• Configure with Gemini (key + model) and verify a search query returns results
• Configure with OpenAI-compatible (key + base URL + model) and verify a search query returns results
• Verify /aisearch/validate succeeds for the selected provider

Notes

• OpenAI-compatible integration uses the Chat Completions endpoint (/v1/chat/completions).