Add alert rules for load

[facundofc]

Enhancement Proposal

Much like what's mentioned on #12 but I didn't find a way to reopen the issue.

We are having an issue in one of our k8s clusters in which processes in the D state begin to accumulate, rendering pods and other services running on the affected worker unresponsive. We didn't get any HostHigh*WaitingTime alert, because they actually don't fire. However, the worker's load does spike abruptly. I'm attaching screenshots of the situation.

It seems that the HostHigh*WaitingTime alert rules alone are not enough to cover all scenarios. We might need to monitor load after all.