feat: add MiniMax provider support for prompt rewriting

[octo-patch]

Summary

This PR adds MiniMax as a supported provider for the prompt rewriting feature in JoyAI-Image.

Changes

• src/infer_runtime/settings.py: Auto-detects MiniMax models (any model whose name starts with minimax, case-insensitive) and configures MINIMAX_API_KEY and the MiniMax OpenAI-compatible base URL (https://api.minimax.io/v1) automatically.
• src/infer_runtime/prompt_rewrite.py: Fixes temperature handling for MiniMax — MiniMax requires temperature in (0.0, 1.0] and rejects 0.0, so MiniMax models now default to 1.0.
• inference.py: Updated --rewrite-model help text to document MiniMax support.
• tests/test_minimax_support.py: Unit tests (20) + integration tests (2, skipped without MINIMAX_API_KEY).

Usage

# Set your MiniMax API key
export MINIMAX_API_KEY=your-minimax-api-key

# Use MiniMax for prompt rewriting
python inference.py \
  --ckpt-root /path/to/ckpts \
  --prompt "Turn the plate blue" \
  --image test_images/test_1.jpg \
  --rewrite-prompt \
  --rewrite-model MiniMax-M2.7

When --rewrite-model is set to a MiniMax model (e.g. MiniMax-M2.7 or MiniMax-M2.7-highspeed), the settings layer automatically:

1. Reads MINIMAX_API_KEY from the environment
2. Sets the base URL to https://api.minimax.io/v1 (unless OPENAI_BASE_URL is already set)

Supported Models

Model	Description
MiniMax-M2.7	Default — peak performance
MiniMax-M2.7-highspeed	Faster, same performance

API Reference

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