Add MiniMax as built-in LLM provider

[octo-patch]

Summary

• Add MiniMax-M2.5 and MiniMax-M2.5-highspeed as built-in default models via new minimax_models.py default plugin
• Reuse existing OpenAI-compatible Chat/AsyncChat classes with MiniMax-specific API base URL and key management
• Enforce MiniMax temperature constraint (0, 1] via custom MiniMaxOptions (MiniMax API rejects temperature=0)
• Register model aliases: minimax, m2.5, minimax-fast, m2.5-highspeed

Usage

llm keys set minimax
# Paste MiniMax API key here

llm -m MiniMax-M2.5 "Tell me about the Great Wall of China"
llm -m minimax-fast "Summarize this text" < article.txt

Changes

File	Description
llm/default_plugins/minimax_models.py	New default plugin registering MiniMax models
llm/plugins.py	Added minimax_models to DEFAULT_PLUGINS
tests/test_minimax_models.py	30 tests (27 unit + 3 integration)
README.md	Added MiniMax to provider list and quick start
pyproject.toml	Added MiniMax to description

Test plan

• 27 unit tests pass (model registration, options validation, mocked HTTP streaming/non-streaming, usage tracking, aliases)
• 3 integration tests pass against live MiniMax API (skipped when MINIMAX_API_KEY not set)
• Existing test suite unaffected (pre-existing test_chat_tools failure is unrelated)

MiniMax API Details

• Base URL: (OpenAI-compatible)
• Models: MiniMax-M2.5 (204K context), MiniMax-M2.5-highspeed (faster variant)
• API Key: Set via or env var
• Temperature: Must be in (0, 1] — validated by custom Options class