Add an extended resource name for full GPUs requests in the default device-class

[shivamerla]

In the default device class (gpu.nvidia.com) created by the DRA Helm chart for full GPUs, add an extended resource name field corresponding to nvidia.com/gpu. This enables users to continue using extended resource requests in their Pod specs, whether those resources are provided by the device plugin or by DRA. By including the extended resource name, existing workloads can run unchanged, and the scheduler can correctly match Pods to devices associated with the device class based on their extended resource requests.

[shivamerla]

Proposed a change with #737