feat: add MiniMax as alternative LLM provider

[octo-patch]

Summary

• Add utils/llm_provider.py multi-provider utility supporting OpenAI and MiniMax (M2.5 / M2.5-highspeed) via OpenAI-compatible API
• Add tutorial notebook 20.multi-provider-llm-with-minimax.ipynb demonstrating chat, prompt templates, chains, streaming, and side-by-side provider comparison
• Update README with provider documentation and usage examples

What is MiniMax?

MiniMax provides large language models with up to 204K context window. Their API is OpenAI-compatible, making integration straightforward via LangChain's ChatOpenAI.

How it works

Set LLM_PROVIDER=minimax and MINIMAX_API_KEY environment variables to use MiniMax across all notebooks:

from utils.llm_provider import get_chat_model
chat = get_chat_model(provider="minimax", temperature=0)

The utility handles:

• Provider-specific API base URL and model defaults
• Temperature clamping for MiniMax (0.0–1.0)
• Automatic fallback to chat endpoint for providers without completion API

Test plan

• 25 unit tests covering provider resolution, temperature clamping, factory functions
• 3 integration tests with live MiniMax API (chat, system messages, streaming)
• All tests passing

Files changed (7 files, 753 additions)

File	Description
utils/__init__.py	Package init
utils/llm_provider.py	Multi-provider factory (get_chat_model, get_llm, get_embeddings)
20.multi-provider-llm-with-minimax.ipynb	Tutorial notebook
README.md	Provider docs + notebook link
tests/__init__.py	Test package init
tests/test_llm_provider.py	25 unit tests
tests/test_integration.py	3 integration tests