Measure auto-scaled machines teardown duration
Description
In gitlab#215145 (closed) we want to measure the time we spend in a machine creation and a machine teardown when an executor with auto-scaling is being used.
We are currently missing this key metric that will tell us what is the duration and will allow us to calculate histogram quantile in Grafana / Prometheus.
Proposal
Add a Prometheus histogram for docker+machine executor machine teardown.
Links to related issues and merge requests / references
GitLab Runner Auto-scaling v2.0