Clarify the type of container disk and network metrics

[ChrsMark]

From https://www.elastic.co/guide/en/ecs/current/ecs-container.html the container.disk.*bytes and container.network.*.bytes are mentioned to be of type gauge.

However by looking into the code of docker Metricbeat module it seems that those should be defined as counters instead:

From https://github.com/elastic/beats/blob/89bcc33a9a90cbacceaa61fceec0e8d23073574c/metricbeat/module/docker/diskio/data.go#L61 we see that stats.servicedBytes.reads is used to get the value. This seems to be a counter.
This value is also reflected into diskio.read.bytes field which mentions that it is the total number (https://github.com/elastic/beats/blob/2c47ac8d7e3af05dc84c19ae5ff81aeaf3cca875/metricbeat/module/docker/diskio/_meta/fields.yml#L20) of bytes for the life of the container.

Same happens with the network related fields.

I performed some tests to check that these fields are only getting increased:

If those findings are valid we need to:

1. Tune ECS accordingly to not mistakenly mention that those fields are gauges.
2. Revisit the containerd implementation and verify that we emit counter values there as well: https://github.com/elastic/beats/blob/89bcc33a9a90cbacceaa61fceec0e8d23073574c/x-pack/metricbeat/module/containerd/blkio/blkio.go#L45

cc: @kaiyan-sheng @MichaelKatsoulis @mlunadia

[ChrsMark]

By looking into various resources (with @MichaelKatsoulis ) we conclude that those should be defined as counters:

1. For diskio: https://docs.docker.com/config/containers/runmetrics/#block-io-metrics and https://www.kernel.org/doc/Documentation/cgroup-v1/blkio-controller.txt
2. For network it seems that the API serves the values from the TX/RX counters: https://docs.docker.com/config/containers/runmetrics/#interface-level-counters

cc: @elastic/obs-cloudnative-monitoring