feat(ai): add MiniMax as third LLM provider

[octo-patch]

Summary

Adds MiniMax as a third cloud LLM provider alongside Google Gemini and Ollama. MiniMax offers an OpenAI-compatible API, making integration clean via the existing openai SDK.

• New provider: packages/ai/src/services/minimax.ts — full AIModel interface implementation with think-tag stripping and response_format: json_object for structured outputs
• Router update: modelRouter.ts now selects MiniMax when USE_MINIMAX=true and MINIMAX_API_KEY is set
• Environment config: MINIMAX_API_KEY, USE_MINIMAX, MINIMAX_MODEL (defaults to MiniMax-M1)
• Tests: 14 unit tests + 3 integration tests for MiniMax provider, plus MiniMax selection test in model router
• Docs: Updated .env.example and README.md with MiniMax configuration
• Test fix: Fixed vitest alias to resolve @shared from source (was pointing to non-existent dist/), fixed model router test mock path

Configuration

# MiniMax Option
MINIMAX_API_KEY="your-api-key"
USE_MINIMAX="true"
MINIMAX_MODEL="MiniMax-M1"  # or MiniMax-M1-highspeed

Test Plan

• All 14 MiniMax unit tests pass (mock-based)
• All 3 MiniMax integration tests pass (live API)
• All 10 existing model router tests pass (including new MiniMax selection test)
• Provider selection priority: Ollama > MiniMax > Gemini
• Manual test with Docker Compose setup

[IliasHad]

@octo-patch Thank you for your contribution, much appreciated. Would you like to share your feedback on using Mini Max vs Gemini 2.5 Pro or 3.1 Pro? Why did you choose Mini Max specifically, and what feedback about contributing to this project (I wanna make sure that it's easy for other people to contribute)?

PS: I'll review the code once it passes all the GitHub workflows