Enable Non-TEE GPU Inference

[jcabrero]

This task aims to create an environment within AWS / GCP where we have access to a GPU other than an H100 while giving CPU TEE security guarantees. This aims to prove that such an extended environment is not required for reduced inference times.

[jcabrero]

I have checked in the two current available providers AWS and GCP:

• GCP does not offer the possibility of running a GPU node with CPU TEE support.
• AWS does not offer either the possibility of running a node with GPU and TEE support

[jcabrero]

Taking this task back.