Run ComfyUI on AWS Inferentia2

Hi,

AWS has a CPU-based instance type called **Inf2** ([link](https://aws.amazon.com/ec2/instance-types/inf2/)).

From [this](https://aws.amazon.com/it/blogs/machine-learning/maximize-stable-diffusion-performance-and-lower-inference-costs-with-aws-inferentia2/) article, they seem to perform almost as well as other GPU-based instances, but at a lower cost.

We did a quick test running a workflow on an Inf2 instance "as is", but of course is much slower than a GPU-based instance. A particular AWS SDK called [Neurons](https://awsdocs-neuron.readthedocs-hosted.com/en/latest/index.html) is required to compile models so they can run on Inf2 instances at maximum performance.

How ComfyUI can be run on this type of instance? Has anyone done this yet? Or is this a feature that will be included in the future releases of ComfyUI, maybe?

Thanks.