Pytorch plugin e2e probabilistic failure

[wangyang0616]

What happened:
The pytorch-plugin e2e fails probabilistically.

During the preliminary analysis, sometimes the mirror image: volcano/pytorch-mnist-v1beta1-9ee8fda-example:0.0.1 fails to download, causing the use case to fail.

The error message is as follows：

Will run 32 of 32 specs
••••
------------------------------
• [FAILED] [675.825 seconds]
Pytorch Plugin E2E Test [It] will run and complete finally
/home/runner/work/volcano/volcano/test/e2e/jobseq/pytorch_plugin.go:12

  Timeline >>
  STEP: Initializing test context @ 04/11/23 12:51:24.026
  STEP: Creating Queues @ 04/11/23 12:51:24.228
  STEP: Cleaning up test context @ 04/11/23 13:02:22.847
  [FAILED] in [It] - /home/runner/work/volcano/volcano/test/e2e/jobseq/pytorch_plugin.go:56 @ 04/11/23 13:02:39.85
  << Timeline

  [FAILED] Unexpected error:
      <*errors.errorString | 0xc00001c120>: {
          s: "[Wait time out]: expected job 'pytorch-job' to be in status Completed, actual get Running",
      }
      [Wait time out]: expected job 'pytorch-job' to be in status Completed, actual get Running
  occurred
  In [It] at: /home/runner/work/volcano/volcano/test/e2e/jobseq/pytorch_plugin.go:56 @ 04/11/23 13:02:39.85
------------------------------
•••••••••••••••••••••••••••

Summarizing 1 Failure:
  [FAIL] Pytorch Plugin E2E Test [It] will run and complete finally
  /home/runner/work/volcano/volcano/test/e2e/jobseq/pytorch_plugin.go:56

Ran 32 of 32 Specs in 1693.664 seconds
FAIL! -- 31 Passed | 1 Failed | 0 Pending | 0 Skipped
--- FAIL: TestE2E (1693.67s)
FAIL

Ginkgo ran 1 suite in 30m14.639645[387](https://github.com/volcano-sh/volcano/actions/runs/4667599932/jobs/8263602086?pr=2775#step:7:388)s

What you expected to happen:
CI use case runs run successfully every time.

How to reproduce it (as minimally and precisely as possible):
make e2e

Anything else we need to know?:

Environment:

• Volcano Version: master
• Kubernetes version (use kubectl version): k8s v1.25
• Cloud provider or hardware configuration:
• OS (e.g. from /etc/os-release):
• Kernel (e.g. uname -a):
• Install tools:
• Others:

[wangyang0616]

/assign