Inconsistency in description of perf metric naming between docs and Prometheus export

[bboreham]

In https://github.com/google/cadvisor/blob/5adb1c3bb38b4c5d50b31f39faf3214a44ae479b/docs/runtime_options.md#perf-events, it says:

Event name should be a human readable string that will become a metric name.

As far as I can see, when exported to Prometheus all these metrics have the name container_perf_events_total, and the string becomes the value of a label event.

Maybe the context refers to a different kind of metric name?