Introduce standalone data stream support, deprecate Elasticsearch setup

[axw]

APM Server is currently responsible for installing Elasticsearch (legacy) index templates, ingest pipelines, and ILM policies. We now also have the Fleet integration which installs variants of these, with the index templates creating traces-*, logs-*, and metrics-* data streams instead of apm-* indices.

We intend to increasingly rely on Fleet and Elastic Agent for powering features such as APM agent central config and source map distribution. In the future, we will remove the ability to write to apm-* indices and always create data streams.

We should introduce an option for standalone APM Server to start writing data streams, relying on Fleet (Kibana) to set up the integration package. When this option is enabled, APM Server will wait for the integration package to be installed before writing any data, and will write to the data stream names rather than legacy index names.

I propose that we introduce the option no later than 7.last, where it will be off by default. At the same time we will deprecate Elasticsearch setup.
In 8.0 we will remove the ability to setup Elasticsearch, and turn the option on by default (#5961). It should still be possible to disable the option for scenarios where APM Server lacks privileges to check the package is installed; in this case the user is responsible for ensuring the package is installed before APM Server is started.

[axw]

Related: #5949