Ability to schedule notebooks on Inferentia and Trainium devices

[surajkota]

Is your feature request related to a problem? Please describe.
AWS specific hardware configuration should be available by default in jupyter webapp to be able to run notebook servers on Inferentia and Trainium instance types, e.g.

Describe the solution you'd like
Make Neuroncores device related configurable in Kubeflow notebooks
https://awsdocs-neuron.readthedocs-hosted.com/en/latest/containers/index.html

Vendor information can be exposed in the webapp using spawner_ui config loaded to jupyter-web-app configmap in kubeflow namespace with the Neuron related information

aws.amazon.com/neuron Resources is easy to modify in aws manifests, additional configuration needed is specifying schedulerName which can be added as a feature to PodDefaults or also in Jupyter webapp. This requires an upstream configuration