feat: add MiniMax provider (M3 default) to RAG templates

[octo-patch]

Summary

This PR adds MiniMax as an alternative LLM provider option to the Pathway RAG templates, with MiniMax-M3 (the latest model) as the default example.

MiniMax offers an OpenAI-compatible API at https://api.minimax.io/v1, which means it can be used as a drop-in replacement for OpenAI models via pw.xpacks.llm.llms.OpenAIChat by setting base_url and api_key.

Changes

• Added commented MiniMax configuration examples to app.yaml in adaptive_rag, question_answering_rag, multimodal_rag, and slides_ai_search templates (using MiniMax-M3 as the default model)
• Added MINIMAX_API_KEY to the corresponding .env.example files
• MiniMax-M2.7 and MiniMax-M2.7-highspeed remain available as alternative model IDs users can plug in

How to use MiniMax

Uncomment the MiniMax section in any template's app.yaml and set MINIMAX_API_KEY in your .env file. By default the example uses MiniMax-M3 (512K context, up to 128K output, image input supported); swap the model field to MiniMax-M2.7 if you prefer the previous generation.

Note: MiniMax requires temperature to be in the range (0.0, 1.0].

API Reference: https://platform.minimax.io/docs/api-reference/text-openai-api

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}