[APM] Use ML plugin APIs for querying and capabilities in Service Maps

[ogupte]

For Service map integration with machine learning, APM should list ml in optionalPlugins in kibana.json in order to leverage the mlClient and mlCapabilities APIs exposed by ML's plugin setup contract. Similar to how SEIM handles ml integration (https://github.com/elastic/kibana/blob/master/x-pack/plugins/siem/server/lib/machine_learning/authz.ts#L117-L118). In other areas of integration with ML, this would allow more appropriate messaging in the ml integration flyout, and transaction charts.

[elasticmachine]

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

[jgowdyelastic]

I addition to using mlCapabilities from ML's shared services, all plugins which perform searches on ML's results index will need to switch using our shared mlSearch function which includes license and user capabilities checks. This change is needed for ML's adoption of spaces, where all access to ml indices will need to be secured behind a kibana api.
mlSearch is a basic wrapper around callAsCurrentUser('search'.... which hardcodes the ml results index, so can be called in the way.

[ogupte]

This was accomplished in #68485

[jgowdyelastic]

This may have been closed prematurely as there are still some APM functions which search the ML results index directly.
On master I can still see functions making use of getMlIndex and searching directly.
getMlBucketSize
getAnomaliesData
anomalySeriesFetcher

Unless this work is to be covered by a different issue? The title suggests that it will cover all queries for ML data.

[sorenlouv]

@ogupte Will you create a new issue to track the remaining work?

[ogupte]

I'll change the name and description on this issue, and move remaining work to the new issue

[ogupte]

#69092