Update alerting information in README

[boosty]

The previously described Prometheus approach is not good enough IMO, as it's missing too many OOM events.

[0robustus1]

We could also entertain the idea of exposing this as a metric from the kubernetes-oom-event-generator itself. It would be able to expose these OOM-Kills on a per pod/container basis as a prometheus metric.

[boosty]

@0robustus1 That's an interesting idea, although it overlaps a bit with kube_pod_container_status_last_terminated_reason, which is a gauge.

I think we could indeed provide a more useful metric if we provide it as a counter instead.

See also the related discussion here: kubernetes/kubernetes#69676 (comment)