[Improvement] kbcli support requesting gpu resources

[lynnleelhl]

llm with vllm framework need gpu to run, kbcli should support setting gpu resource request in command line, here's the gpu request example in yaml

apiVersion: apps.kubeblocks.io/v1alpha1
kind: Cluster
metadata:
  name: bc
spec:
  clusterDefinitionRef: vllm # ref clusterdefinition.name
  clusterVersionRef: vllm-baichuan2-13b
  terminationPolicy: Delete
  componentSpecs:
  - name: vllm
    componentDefRef: vllm
    monitor: false
    replicas: 2
    resources:
      requests:
        nvidia.com/gpu: 1
      limits:
        nvidia.com/gpu: 1

[github-actions]

This issue has been marked as stale because it has been open for 30 days with no activity