feat: add MiniMax as first-class LLM provider

[octo-patch]

Summary

This PR adds MiniMax as a built-in LLM provider for CAI, enabling users to leverage MiniMax's large language models (MiniMax-M2.7, MiniMax-M2.5, MiniMax-M2.5-highspeed) via an OpenAI-compatible API.

Changes

• MiniMaxProvider (src/cai/sdk/agents/models/minimax_provider.py): New ModelProvider implementation that creates OpenAIChatCompletionsModel instances configured for MiniMax's API. Reads MINIMAX_API_KEY from environment by default, supports custom API key, base URL, or pre-configured AsyncOpenAI client.
• MiniMax model routing (src/cai/sdk/agents/models/openai_chatcompletions.py): Added MiniMax detection in _fetch_response to correctly route requests through litellm with proper api_base and custom_llm_provider settings.
• Exports (src/cai/sdk/agents/__init__.py): MiniMaxProvider is exported from the agents package.
• Documentation (docs/providers/minimax.md): Setup guide with three usage patterns (MiniMaxProvider, environment variables with LiteLLM, direct model on Agent).
• Example (examples/model_providers/minimax_example.py): Working example demonstrating MiniMax-M2.7 and MiniMax-M2.5-highspeed usage.
• README: Added MiniMax to the multi-model support provider list.

Tests

• 18 unit tests (tests/core/test_minimax_provider.py): Cover initialization, API key handling, env var detection, model resolution, lazy client loading, ModelProvider interface compliance, and Runner integration.
• 3 integration tests (tests/integration/test_minimax_integration.py): Validate real API calls to MiniMax-M2.7, MiniMax-M2.5, and MiniMax-M2.5-highspeed (require MINIMAX_API_KEY).

Available Models

Model	Context Window	Description
MiniMax-M2.7	1M tokens	Latest and most capable model
MiniMax-M2.5	1M tokens	Strong general-purpose model
MiniMax-M2.5-highspeed	204K tokens	Optimized for speed

Usage

from cai.sdk.agents import Agent, Runner, RunConfig
from cai.sdk.agents.models.minimax_provider import MiniMaxProvider

provider = MiniMaxProvider()  # reads MINIMAX_API_KEY from env
agent = Agent(name="assistant", instructions="You are helpful.", model="MiniMax-M2.7")
result = await Runner.run(agent, "Hello!", run_config=RunConfig(model_provider=provider))

Test Plan

• 18 unit tests pass
• 3 integration tests pass with live MiniMax API
• Existing tests unaffected