Add IPEX-LLM section (for local model deployment on Intel CPU/GPU) to tutorial "Connecting with Different Models"

[shane-huang]

This PR adds IPEX-LLM section as a new local model deployment option to the tutorial Connecting with Different Models.

IPEX-LLM is a library which allows users to run LLMs locally on Intel CPU and GPU with very low latency. With IPEX-LLM, user can leverage even the integrated GPU of low-cost PCs to run LLMs with a smooth experience.

dify can successfully connect to local LLMs deployed with IPEX-LLM, and there's a detailed quickstart guide for how to use dify with IPEX-LLM with a demo video at the IPEX-LLM docs website (see Run Dify with IPEX-LLM QuickStart Guide ).

It's the first time I contribute to dify docs. Any suggestions on styles or contents, please kindly let me know.

[crazywoola]

Hello, sorry for the late response.

I read this doc

Regarding the deployment of dify itself, I notice it will start a dev environment instead of production environment. I suggest replacing this part to use docker ways which should have better performances for both frontend and backend part. And there are some conflicts as well.

[crazywoola]

See comments

[crazywoola]

Hello this pr has conflicts that must be resolved, please use the latest main branch to reopen an pr. Thanks :)

[shane-huang]

We will follow-up on your comments and reopen a PR based on latest main branch.