docs: add MiniMax as LLM provider

[octo-patch]

Summary

Adds MiniMax as a supported LLM provider in AutoAgent. MiniMax exposes an OpenAI-compatible Chat Completions endpoint, so we can plug it into LiteLLM by routing through the existing openai/* provider with a custom API_BASE_URL, no library changes required.

Changes

• .env.template: add MINIMAX_API_KEY= placeholder alongside the other optional provider keys.
• README.md: add a MiniMax section under "auto main with different LLM Providers", following the same template as the Grok / OpenAI-compatible entry.

Why

MiniMax-M2.7 and MiniMax-M2.7-highspeed are competitive flagship models with a 200K context window and strong tool-use/coding performance, and the OpenAI-compatible endpoint makes them easy to use from any LiteLLM-based agent. Documenting the right env var + base URL combo saves users from guessing.

Testing

Verified end-to-end via LiteLLM with the documented configuration:

import litellm, os
os.environ["OPENAI_API_KEY"] = os.environ["MINIMAX_API_KEY"]
resp = litellm.completion(
    model="openai/MiniMax-M2.7",
    messages=[{"role": "user", "content": "Reply with the word: pong"}],
    api_base="https://api.minimax.io/v1",
    max_tokens=200,
    temperature=1.0,
)
# -> "pong"

Returns a valid completion against https://api.minimax.io/v1/chat/completions, confirming the documented commands work out of the box.