feat: add MiniMax as alternative LLM provider

[octo-patch]

Summary

Add MiniMax M2.7 and M2.7-highspeed (204K context) as first-class LLM providers across the Medical-Graph-RAG pipeline:

• Project-level code (utils.py, summerize.py): Provider auto-detection via MINIMAX_API_KEY env var, LLM_PROVIDER / LLM_MODEL overrides, temperature clamping to MiniMax (0.0, 1.0] range
• nano_graphrag (_llm.py): Async minimax_complete_if_cache(), minimax_m27_complete(), and minimax_m27_highspeed_complete() functions with caching and temperature clamping
• CAMEL framework: MiniMaxModel backend class, MiniMaxConfig, ModelType.MINIMAX_M27/M27_HIGHSPEED enums, ModelPlatformType.MINIMAX, ModelFactory registration
• README: MiniMax usage docs with examples for all three integration layers

Usage

export MINIMAX_API_KEY=your_key
# Auto-detected from API key, or set explicitly:
export LLM_PROVIDER=minimax
python run.py -simple True

Note: MiniMax does not have a public embedding API, so embeddings fall back to OpenAI text-embedding-3-small when using MiniMax as the LLM provider.

Changes

File	Change
utils.py	Provider presets, auto-detection, temp clamping
summerize.py	Provider-aware call_openai_api()
nano_graphrag/_llm.py	MiniMax async completion functions
camel/types/enums.py	ModelType.MINIMAX_M27/M27_HIGHSPEED, ModelPlatformType.MINIMAX
camel/configs/minimax_config.py	MiniMaxConfig + MINIMAX_API_PARAMS
camel/models/minimax_model.py	MiniMaxModel backend with temp clamping
camel/models/model_factory.py	MiniMax branch in ModelFactory.create()
camel/models/init.py	Export MiniMaxModel
camel/configs/init.py	Export MiniMaxConfig, MINIMAX_API_PARAMS
README.md	MiniMax setup docs + code examples
tests/test_minimax_unit.py	37 unit tests
tests/test_minimax_integration.py	4 integration tests

12 files changed, ~842 additions

Test plan

• 37 unit tests pass (provider detection, temp clamping, enums, config, model backend, factory, nano_graphrag functions)
• 4 integration tests pass against live MiniMax API (utils call_llm, nano_graphrag M2.7/M2.7-highspeed, CAMEL ModelFactory)
• Verify existing OpenAI-based pipeline still works unchanged (backward compatible)