Add MiniMax as first-class LLM provider with auto-detection

[octo-patch]

Summary

• Add MiniMax as a first-class LLM provider with auto-detection
• When MINIMAX_API_KEY is set, the provider is auto-configured — no need to set LLM_PROVIDER, model, or base URL manually
• Built-in defaults: MiniMax-M2.5 model, https://api.minimax.io base URL
• Temperature clamping to MiniMax's accepted range (0.01–1.0)
• Users can override defaults via MINIMAX_MODEL and MINIMAX_BASE_URL env vars
• Provider defaults system (_PROVIDER_DEFAULTS) is extensible for future providers

Changes

File	Change
utils/call_llm.py	Auto-detect MiniMax via API key, provider defaults dict, temperature clamping
.env.sample	MiniMax configuration example
README.md	Document MiniMax as a supported provider
tests/test_call_llm.py	15 unit tests + 3 live integration tests
tests/__init__.py	Test package init

Usage

# Just set the API key — everything else is auto-configured
export MINIMAX_API_KEY="your-key"
python main.py --repo https://github.com/username/repo

# Or override model/base URL
export MINIMAX_MODEL="MiniMax-M2.5-highspeed"

Test plan

• 15 unit tests pass (provider detection, defaults, request construction, caching)
• 3 live integration tests pass (simple completion, multi-line response, highspeed model)
• Existing Gemini/generic provider behavior unchanged