feat: add MiniMax prompt enhancer node

[Octopus (octo-patch)]

Summary

• Add MiniMaxPromptEnhancer ComfyUI node that calls the MiniMax chat completions API to cinematically enhance LTX-Video generation prompts
• Supports both T2V (text-to-video) and I2V (image-to-video) cinematic enhancement modes, reusing the existing system prompts from prompt_enhancer_utils.py
• Supports MiniMax-M2.7 (default) and MiniMax-M2.7-highspeed models
• Automatically strips <think>…</think> reasoning tokens that MiniMax-M2.7 may emit
• An API-based alternative to the local LLM prompt enhancer — no local GPU / VRAM required

Why

The existing LTXVPromptEnhancerLoader + LTXVPromptEnhancer nodes require downloading and running Llama 3.2 locally (several GB, significant VRAM). This new node lets users get high-quality cinematic prompt enhancement via the MiniMax API without any local model loading, similar in spirit to how GemmaAPITextEncode provides an API alternative to local Gemma.

Usage

1. Set your MINIMAX_API_KEY (get one at https://platform.minimax.io/)
2. Add the 🅛🅣🅧 MiniMax Prompt Enhancer node to your ComfyUI workflow
3. Connect its enhanced_prompt output to a CLIP text encode node

API Reference

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

Test plan

• 20 unit tests pass (python -m pytest tests/)
• Integration test against MiniMax API verified locally
• Node appears in ComfyUI under api node/text/Lightricks category
• Registered in both NODE_CLASS_MAPPINGS and nodes_registry.py