[ML] AIOps Log Rate Analysis: Allow the baseline selection window to be set after the deviation window.

[face0b1101]

Describe the feature:
In AIOps Log Rate Analysis, allow the baseline selection to be set after the deviation. Currently, the baseline data window can only be selected prior to the deviation window.

Describe a specific use case for the feature:
When investigating a log rate spike that occurs early in a dataset, and there is no adequate baseline/BAU data prior to the log spike.

The screenshot above illustrates an example where the spikes I am interested in occur before the baseline data. In this example I would like baseline and deviation to be swapped. There is no additional data available in this dataset prior to the spikes, so the baseline data is only available after the deviation.

[elasticmachine]

Pinging @elastic/ml-ui (:ml)

[walterra]

We recently merged support for analysing log rate dips, this should fit this use case since effectively analysing dips swaps baseline and deviation: You will see which items are statistically significant in the baseline time range and are less in numbers missing from the deviation time range.

Closing this for now, please feel free to reopen should dip support not fully solve your use case.