workaround VM first start failure due to swtpm

[chunfuwen]

On rhel 10, swtpm will fail VM first start due to existed known issue, so here it is just to start vm again once first start failure

[chunfuwen]

tpm_device.normal_test.tpm-crb_model.emulator.encrypted.pcrbank_all.restart_libvirtd 49 sec Fixed
tpm_device.normal_test.tpm-crb_model.emulator.encrypted.pcrbank_all.restart_vm.remove_pcrbank 1 min 12 sec Fixed
tpm_device.normal_test.tpm-crb_model.emulator.encrypted.pcrbank_all.restore_vm 1 min 4 sec Fixed
tpm_device.normal_test.tpm-crb_model.emulator.encrypted.pcrbank_all.snapshot_operate 56 sec Fixed
tpm_device.normal_test.tpm-crb_model.emulator.encrypted.pcrbank_all.suspend_resume 52 sec Fixed
tpm_device.normal_test.tpm-crb_model.emulator.encrypted.pcrbank_all.undefine_create 1 min 19 sec Fixed
tpm_device.normal_test.tpm-crb_model.emulator.encrypted.pcrbank_default.basic 45 sec Fixed

[chunfuwen]

@yanqzhan since existed swtpm issue fail lots of cases, we may need this workaround solution.
Once issue is really resolved, we can remove these code accordingly.

[yanqzhan]

Looks good, thank for this. but
vm2 also needs it:
L708 ret = virsh.start(vm2_name, ignore_status=True, debug=True)

IMO, suggest to add "CI_PR": "tp-libvirt xxxx" in ci jobs.yaml instead of merge here and remove later.

[yanqzhan]

LGTM now.