[Logs UI] Allow users to specify datasets in ML job configuration

[mukeshelastic]

ℹ️ This has been split out of #59005.

Summary

Today at ML job configuration, we allow users to specify index to choose from. With this feature, we will allow them to choose a specific dataset as well. This helps users to have granular control on running ML jobs which can be useful when users are trying to reserve ML capacity on the datasets that matter the most as well as filter out datasets not suitable for categorization.

Mockup

Please note that showing the warning indicator is not part of this issue and will be handled in #61900.

[elasticmachine]

Pinging @elastic/logs-metrics-ui (Team:logs-metrics-ui)

[jgowdyelastic]

Could you please expand on what constitutes a "dataset" in this situation?
Is this something that could be achieved by overriding the query used by the job's datafeed?

[weltenwort]

The dataset is determined by the keyword field event.dataset. Our plan is indeed to implement it via a filter in the datafeed, so it's purely a change on our side.