[REQ] auto detect runtime for an inference container

[sozercan]

What kind of request is this?

Improvement of existing experience

What is your request or suggestion?

https://github.com/mudler/LocalAI/issues/1888

Are you willing to submit PRs to contribute to this feature request?

• [ ] Yes, I am willing to implement it.

[sozercan]

docker already has this functionality built-in with platforms (such as linux/amd64/v2, linux/amd64/v3, etc) https://en.wikipedia.org/wiki/X86-64#Microarchitecture_levels

This doesn't exist for gpus

[sozercan]

fixed with #225