Fix CPU usage metrics

[SuperQ]

The CPU usage metrics from ECS are pass-through of Docker stats API.

• Switch the API json struct to use upstream Docker types.
• The API specifies the units are in nanoseconds, not jiffies.

Fixes: https://github.com/prometheus-community/ecs_exporter/issues/34

Signed-off-by: SuperQ superq@gmail.com

[jseiser]

@SuperQ

Any chance you can get this merged/built?

Thanks.