funcx kubernetes endpoint workers can have duplicate pod names

[benclifford]

Describe the bug
Sometimes the endpoint running in kubernetes will generate worker pods with the same name.

For example:

# minikube kubectl get pods
NAME                                            READY   STATUS    RESTARTS        AGE
funcx-1632329996841                             1/1     Running   0               4m3s
funcx-1632329998841                             1/1     Running   0               4m2s
funcx-1632330007842                             1/1     Running   0               3m53s
funcx-endpoint-86756c48c8-nc7r2                 1/1     Running   2 (10m ago)     12m

This is because the pods are named after the current millisecond, which is not a unique ID.

This makes referring to those pods awkward on the commandline.

crossref this bug this is forked from: Parsl/parsl#1122

To Reproduce

Get the endpoint to launch two pods in the same millisecond.

Expected behavior
Distinct pods should have distinct names.

Environment
my own minikube on ubuntu

[rjmello]

The forked KubernetesProvider is now deprecated, and Parsl/parsl#3639 addresses this issue for Parsl's KubernetesProvider.