Instrument Fleet's setup APIs with APM spans

[joshdover]

Related to https://github.com/elastic/dev/issues/1443

We'd like to keep an eye on the performance of Fleet's setup sequence to ensure that we do not degrade the performance of this over time and to help spot any performance optimizations that may be available today. The performance of this process is important to ensure a smooth onboarding process for new users.

Similar to #109055, we'd like to be able to monitor how the performance of this API changes over time and to alert us if any major performance regressions are introduced. Kibana's CI already enables and collects APM data during executions of the functional test suite, and each HTTP route is already instrumented as a transaction. We should be able to leverage the existing data to establish a baseline. Adding custom APM spans would allows us to better identify which parts of the setup process are taking the most time.

Cloud's integration testing should also be able to leverage APM to measure this over time and ensure that Fleet's setup time is also staying performant in that environment.

[elasticmachine]

Pinging @elastic/fleet (Team:Fleet)

[lizozom]

Is this a subset of #129050?

As for the infrastructure, if we want to do this in CI, this sounds like it could be a follow up phase to https://github.com/elastic/kibana-team/issues/501?