[kube-prometheus-stack] Keep KubeletDown cluster-scoped

[Zakharden]

What this PR does / why we need it

KubeletDown joins kube_node_info and kubelet up{} series to detect missing kubelet targets. When defaultRules.additionalAggregationLabels includes labels that do not exist consistently on both metric sources, such as namespace, the alert can fire even when kubelets are healthy.

This keeps KubeletDown scoped to cluster only, while leaving the existing additional aggregation label behavior unchanged for other generated rules.

Which issue this PR fixes

Fixes #6872.

Special notes for your reviewer

The generated rules file is updated through hack/sync_prometheus_rules.py; the follow-up commit only syncs the generated file formatting expected by CI.

Checklist

• DCO signed
• Chart Version bumped
• Title of the PR starts with chart name

Validation

• helm unittest -f unittests/prometheus/kubelet_down_aggregation_test.yaml charts/kube-prometheus-stack
• ./venv/bin/python ./sync_prometheus_rules.py from charts/kube-prometheus-stack/hack
• git diff --check

[Zakharden]

Hi maintainers, could someone take a look at this PR?

It keeps KubeletDown cluster-scoped so defaultRules.additionalAggregationLabels do not introduce false positives when labels like namespace are not present consistently across kube_node_info and kubelet up{}.

I synced the generated rules and added focused unittest coverage; happy to revise if there is a preferred rule-generation pattern. Thanks.

@jkroepke

[Zakharden]

@jkroepke @gkarthiks @Xtigyro @gianrubio Rebased on the latest main; the PR is mergeable now and CI is green. I also updated the unit test to assert the KubeletDown rule by content rather than by a brittle array index. Could you please take another look?