Add time_series_metric parameter

[csoulios]

This PR adds the time_series_metric parameter to the following field types:

• Numeric field types
• histogram
• aggregate_metric_double

The time_series_metric parameter is of type string (default is null) and is used to mark that a field is a time series metric field with a specific metric type. The following metric types are supported for each Elasticsearch field type:

Metric type	ES field type
gauge	number, aggregate_metric_double
counter	number, aggregate_metric_double
histogram	histogram
summary	aggregate_metric_double

Example of setting the time_series_metric parameter to a field can be seen here:

PUT my-index-000001
{
  "mappings": {
    "properties": {
      "free_memory": {
        "type": "double",
        "time_series_metric": "gauge"
      },
      "net_bytes_in": {
        "type": "double",
        "time_series_metric": "counter"
      }
    }
  }
}

This PR together with #74450 annotate which Elasticsearch fields are time series dimensions or metrics.

Closes #74014

[elasticmachine]

Pinging @elastic/es-search (Team:Search)

[elasticmachine]

Pinging @elastic/es-analytics-geo (Team:Analytics)

[imotov]

It looks good to me in general. I have only a couple of comments. I think some integration level tests, maybe even rest test could be helpful, so we have some clear examples how this is used. It also looks like long fields can be metrics and dimensions at the same time, not sure if this is intentional. It would be great if @nik9000 could review it since he is working with this at the moment.

[nik9000]

It would be great if @nik9000 could review it since he is working with this at the moment.

I'm not 100% sure we'll keep allowing long to be a dimension. But I guess it's easy enough to add the assertion that you can't do that so we may as well.

[imotov]

We will not keep it as part of routing... but nothing prevents us from keeping it as non-routing dimension.

[nik9000]

LGTM.

Sorry for the delay!