[Logs UI] Add analysis tab

[weltenwort]

Summary

After the Logs UI will have gained a top-level tab bar in #39071, the tab for the new ML-based log analysis screen needs to be added there.

Acceptance criteria

• Under certain conditions an "Analysis" tab is present to the right of the "Stream" tab, but to the left of the "Settings" tab.
• The tab is only present if the cluster has the required license and the user has the necessary permissions to use the categorization ML feature.
• The tab's content depends on the existence of the jobs defined by the Machine Learning module (see the implementation notes below):
  • If all the jobs exist, a result screen ([Logs UI] Create screen to show results of the analysis ML jobs #41881) is displayed. For the purpose of this issue, a placeholder can be shown in its place.
  • If not all the jobs exist, a setup screen ([Logs UI] Create screen to set up analysis ML jobs #41877) is displayed. For the purpose of this issue, a placeholder can be shown in its place.
• Failure scenarios
  • failed to get capabilities → hide tab

Implementation notes

• The /api/ml/ml_capabilities API provides details about the user's capabilities. According to @jasonrhodes's research, the capabilities.canCreateJob is probably the best capability to use as a visibility criterion.
• The existence and integrity of the jobs can be determined using the /api/ml/jobs/jobs_summary api, which takes a list of job ids as arguments:
  

  kibana/x-pack/legacy/plugins/ml/server/routes/job_service.js

  Lines 78 to 91 in a9e0f25

  	route({
  	method: 'POST',
  	path: '/api/ml/jobs/jobs_summary',
  	handler(request) {
  	const callWithRequest = callWithRequestFactory(elasticsearchPlugin, request);
  	const { jobsSummary } = jobServiceProvider(callWithRequest);
  	const { jobIds } = request.payload;
  	return jobsSummary(jobIds)
  	.catch(resp => wrapError(resp));
  	},
  	config: {
  	...commonRouteConfig
  	}
  	});

  
  The job ids used by this integration module are deterministically derived from the space id and the source id.

[elasticmachine]

Pinging @elastic/infra-logs-ui

[spong]

Here's how the SIEM team ended up implementing the ml_capabilities checks in case this is of any (re)use :)

[SIEM] Adds Machine Learning Permissions to hide or show UI elements based on the permissions

There's also some custom ML API error parsing for displaying errors in toasts that might be useful as well. Feel free to reach out to myself or @FrankHassanabad as we just went through this for SIEM -- happy to chat about the things we encountered along the way.

[weltenwort]

That's very helpful, thank you!

[Kerry350]

@spong Thanks for the above, that'll be very useful 👌