It's a lot easier to show and explain the Prometheus config if available
in the standard yaml format. The Kubernetes ConfigMap specs can always
be auto-generated.
This also contains the generated ConfigMaps for the grafana dashboards
as well as for alerts. These will be updated in subsequent changes as
sets of alerts and dashboards are added/updated in the respective
subdirectories in `/assets`.
Kubernetes component metrics often have a namespace label themselves,
which is actually relevant. The target namespace label merely informs
where the scrape target is located, which is generally irrelevant for
these specific metrics.
The Prometheus server monitoring Kubernetes should be default cover:
- node exporter
- kubelets
- kube-state-metrics
- apiservers
The version in this commit still has certificate issues with the API
server in minikube. Kubelet scraping likely only works in minikube as it
doesn't seem to use https.
