Description

This PR is to address these issues:

• - Engs find it is very hard to have core or uikit modules update reflect the web app
• - Inference errors are not output to file which is hard to trace the issue
• - Switch between conversations quickly will prompt user a bunch of errors
• - inferenceRequest exported from extension return any value
• - Inference extension should expose more abstracted functions such as stopInference, loadModel(with: parameters), ...
• - Clearer error message to user when they use a wrong Nitro binary (different arch)