feat: add MiniMax as built-in LLM provider

[octo-patch]

Summary

Add MiniMax as a first-class built-in LLM provider, giving users access to MiniMax M2.7, M2.5, and M2.5-highspeed models via the OpenAI-compatible API.

Changes

• Backend (config_service.py): Add MiniMax to DEFAULT_PROVIDERS_CONFIG with API URL and three text models
• Frontend (AddProviderDialog.tsx): Add MiniMax to predefined PROVIDER_OPTIONS dropdown with preset models and API URL
• Frontend (constants.ts): Add MiniMax to PROVIDER_NAME_MAPPING with display name and icon
• README: Mention MiniMax in supported providers list (both EN and ZH)
• Tests: 16 unit tests + 3 integration tests covering config, routing, and API connectivity

How it works

MiniMax provides an OpenAI-compatible API, so it works seamlessly with the existing ChatOpenAI client in agent_service.py. No additional dependencies required. Users just need to:

1. Add MiniMax from the provider dropdown in Settings
2. Enter their MiniMax API key
3. Select a model and start chatting

Test plan

• 16 unit tests pass (python -m pytest tests/test_minimax_provider.py)
• 3 integration tests pass with real API calls (python -m pytest tests/test_minimax_integration.py)
• Manual: add MiniMax provider in Settings UI, verify models appear, send a chat message