APM telemetry for aggregated transactions

[smith]

Add telemetry for aggregated transactions. I'd like to record the number of metric documents we can expect with the current implementation. We can simulate this by requesting data for 1m, adding a composite aggregation on a number of fields and counting the number of buckets. The resulting count will be an approximation of the amount of metric documents that will be created. We should record both the expected metric document count plus the transaction count for that time range.

I'd like to record several combinations:

• current_implementation: observer.name, transaction.name, transaction.result, transaction.type, agent.name, service.environment, service.version, host.hostname, container.id, kubernetes.pod.name, user_agent.name
• no_observer_name: transaction.name, transaction.result, transaction.type, agent.name, service.environment, service.version, host.hostname, container.id, kubernetes.pod.name, user_agent.name
• no_rum: observer.name, transaction.name, transaction.result, transaction.type, agent.name, service.environment, service.version, host.hostname, container.id, kubernetes.pod.name (add a filter to exclude RUM data)
• no_rum_no_observer_name: observer.name, transaction.name, transaction.result, transaction.type, agent.name, service.environment, service.version, host.hostname, container.id, kubernetes.pod.name (add a filter to exclude RUM data)
• only_rum: observer.name, transaction.name, transaction.result, transaction.type, agent.name, service.environment, service.version, host.hostname, container.id, kubernetes.pod.name, user_agent.name (add a filter to exclude all non-RUM data)
• only_rum_no_observer_name: transaction.name, transaction.result, transaction.type, agent.name, service.environment, service.version, host.hostname, container.id, kubernetes.pod.name, user_agent.name (add a filter to exclude all non-RUM data)

I'd also like to record the cardinality of client.geo.country_iso_code separately. (This is done in #71112.)

For an example of how to create the composite aggregation and paginate through the buckets, see scripts/aggregate-latency-metrics/index.ts.

(this description is copied from what @dgieselaar wrote on #71112. Making a separate issue so it can be done in a separate PR.)

[elasticmachine]

Pinging @elastic/apm-ui (Team:apm)