[Logs UI] Add the log categories result screen

[Kerry350]

Note: This issue previously talked about a rare categories section of a log analysis tab. This section has since been split into a separate tab.

Summary

The new categories tab displays the top N categories by count in a given time interval. The interaction (blue button appears on hover) is the same as described for View in context #50304

Acceptance criteria

• The "Categories" tab is shown when the user has the "create ML job" capabilities (same as the log rate tab).
• The tab shows the categorization analysis results in a table, which...
  • is paginated.
  • has a default page size of 25.
  • has the columns "Message count", "Category", "Dataset" and "Max. anomaly score".
  • a sparkline indicates the trend (compared to the previous timeframe) of the "Message count" column.
  • is sorted by message count.
• The dropdown above the table...
  • filters the table rows for those with matching datasets.
  • does not apply any filters when empty.
  • can be easily cleared.
• Failure scenarios:
  • job not set up → handled higher up in the hierarchy by showing the setup screen
  • job configuration outdated → show a callout with a button to enter the setup procedure again
  • job definition outdated → show a callout with a button to enter the setup procedure again
  • no data → show placeholders

Mockups

Figma link: https://www.figma.com/file/T08kdh6YQcVCx1tzljqNHF/Logs-ML-Integration

[elasticmachine]

Pinging @elastic/infra-logs-ui

[katrin-freihofner]

@Kerry350 @weltenwort what do you think about dropping the sparkline if we have too little information (timeframe < 45 min)?

[weltenwort]

Yes, that sounds reasonable. Do we keep the "+N %" in that case?

[weltenwort]

Talking about the sparklines: Does each row's sparkline have its own y-axis maximum or do they share the same value in order to be comparable?