feat: add MiniMax provider support for deep-analysis mode

[octo-patch]

Summary

• Add minimax_provider.py with an OpenAI-compatible client targeting the MiniMax API (international endpoint https://api.minimax.io/v1)
• Support MiniMax-M2.7 (default) and MiniMax-M2.7-highspeed as analysis models
• Add --provider minimax flag to main.py so users can run deep-analysis using MiniMax instead of NotebookLM
• Add optional --model flag for model selection
• Add openai>=1.0.0 to requirements.txt
• Add 17 unit tests in tests/test_minimax_provider.py (all passing)

Usage

# Set API key
export MINIMAX_API_KEY=your_key_here

# Deep-analysis using MiniMax-M2.7 (no NotebookLM auth required)
python main.py /path/to/file.epub --deep-analysis --provider minimax

# Use the faster model
python main.py /path/to/file.md --deep-analysis --provider minimax --model MiniMax-M2.7-highspeed

API References

• Chat (OpenAI Compatible): https://platform.minimax.io/docs/api-reference/text-openai-api

Implementation Notes

• Temperature is fixed at 1.0 per MiniMax API constraints (range: (0.0, 1.0])
• Content is truncated to 50 000 characters when passed as context
• The default NotebookLM workflow is unchanged; --provider minimax opts into the new path