Add dashboards on delay due to controllers downtime

[matte21]

Add dashboards to display the Prometheus metrics added by MikeSpreitzer/kube-examples#119.

For every "implementation action" (e.g. address assignment, local network interface creation, etc...) and (last_client_write, last_controller_start) pair, there's a plot with:

1. Absolute number of actions that suffered a delay.
2. Fraction of delayed actions wrt total actions.
3. Average delay seconds.

The new plots are based on Prometheus histograms but do not use PromQL's rate() function to allow lazy materialization of metrics.