Co-authored with @peytonrunyan
Closes #7739

Converts Kubernetes job PIDs from {cluster_name}:{job_name} to {cluster_uid}:{namespace}:{job_name}. Cluster names can be arbitrarily set in the Kubernetes config and are not a reliable unique identifier. Additionally, the config does not exist when using an in-cluster agent or in cases like #7739. We generate a stable cluster UID from the kube-system namespace UID as described in kubernetes/kubernetes#44954 — which tracks the lack of a Kubernetes API for retrieving a cluster identifier. I attempted to use the job UID instead of the job name because it is guaranteed to be unique across namespaces. However, there is not a clear way to look up a job by its UID later without listing all jobs. Instead, we include the namespace in the PID to ensure we do not kill jobs in the wrong namespace.

Additionally, in #7701 the KubernetesJobResult returns were not updated to include the new identifier. Here, I do some minor refactoring to fix this.

Example

from prefect.infrastructure import KubernetesJob
import anyio


async def main():
   async with anyio.create_task_group() as tg:
       job = KubernetesJob(
           command=["sleep", "1000"], pod_watch_timeout_seconds=5
       )

       pid = await tg.start(job.run)
       print("PID:", pid)

       await job.kill(pid)
       print("Killed :)")


anyio.run(main)

Checklist

• This pull request references any related issue by including "closes <link to issue>"
  • If no issue exists and your change is not a small fix, please create an issue first.
• This pull request includes tests or only affects documentation.
• This pull request includes a label categorizing the change e.g. fix, feature, enhancement