testground/infra: Missing pod(s) in the test run

[Bidon15]

Now, in the EKS cluster, we can rarely see that a set of test pods can be missing out of nowhere

Some sample logs:

Nov 22 17:48:27.938574	�[35mDEBUG�[0m	testplan pods state	{"runner": "cluster:k8s", "run_id": "cdugl9926un8din2mr60", "running_for": "2m2s", "succeeded": 0, "running": 7, "pending": 1, "failed": 0, "unknown": 0}
Nov 22 17:48:29.948165	�[35mDEBUG�[0m	testplan pods state	{"runner": "cluster:k8s", "run_id": "cdugl9926un8din2mr60", "running_for": "2m4s", "succeeded": 0, "running": 8, "pending": 0, "failed": 0, "unknown": 0}
Nov 22 17:48:29.948214	�[34mINFO�[0m	all testplan instances in `Running` state	{"runner": "cluster:k8s", "run_id": "cdugl9926un8din2mr60", "took": "2m4s"}
Nov 22 17:49:22.248158	�[35mDEBUG�[0m	testplan pods state	{"runner": "cluster:k8s", "run_id": "cdugl9926un8din2mr60", "running_for": "2m56s", "succeeded": 0, "running": 8, "pending": 0, "failed": 0, "unknown": 0}
Nov 22 17:49:24.258297	�[35mDEBUG�[0m	testplan pods state	{"runner": "cluster:k8s", "run_id": "cdugl9926un8din2mr60", "running_for": "2m58s", "succeeded": 0, "running": 7, "pending": 0, "failed": 0, "unknown": 0}