[Lens] Support small multiples / facet / trellis for XY and Partition charts

[wylieconlon]

This issue tracks the Lens implementation of the underlying Elastic-charts feature: elastic/elastic-charts#500

The simplest way to integrate this in Lens is by adding a new dimension group to each visualization which allows the user to split the chart by a given aggregation.

[elasticmachine]

Pinging @elastic/kibana-app (Team:KibanaApp)

[wylieconlon]

We've been unblocked on this for a while because of elastic-charts support for small multiples, but we should be able to reuse a lot of the logic from the Visualize implementation that was recently merged: #86880

[flash1293]

Thanks for bringing it up @wylieconlon - I think there are a few things we have to decide on:

• in terms of aggregation nesting order - do we default to keep the "small multiple" dimension the top level dimension at all times? Do we introduce another UI to order the dimensions? Something else?
• in terms of charting capabilities - is this an additional dimension per layer? Is it a new chart type? Is it a configuration for how the single "break down by" dimension is used? do we allow rows and column splits at the same time?

[th0ger]

Will small multiples be the official name for this feature?
This kind of chart is often called facet or trellis (afaik they are the same).
Anyway, now that I've mentioned it, people like me can find this issue :-)

[flash1293]

We haven't decided on the naming in the UI yet, but we will definitely consider your feedback, thanks

[ghudgins]

Related requirements: #2021

[flash1293]

Even with small multiples it’s useful to mix data from different layers in a single chart - for example having a a data series and a threshold series coming from two different places, split by the same field. Each data series should go with their respective threshold series, but multiple data and threshold series in a single chart are unreadable so splitting the chart into rows per data/threshold series makes sense.

[stratoula]

When we start analyzing this it would be good to take under consideration this feature request too #2021