simplify geo_shape geometry for query response

[nreese]

Enhancement request

Allow for simplification of geo_shape geometry in _search response.

Kibana maps use case

In Kibana, the optimal way to retrieve documents is vector tile search API. However, Kibana can not always use the vector tile search API and must fallback to JSON search API under the following circumstances

• Formatted labels
• Data driven styling from join metrics with 'Label', 'Label size', icon 'Symbol size', and 'Symbol orientation' style properties
• Spatial joins (in development)

In cases where JSON search API is required, loading complex geometries is noticeably slow. The performance is effect by having to transmit large amounts of ASCII data across the network. Being able to simplify geo_shape geometry for query response would mitigate the problem by transferring significantly less data across the network.

One risk is that geometry simplification may be more time consuming than transferring the original geometry across the network.

[elasticsearchmachine]

Pinging @elastic/es-analytics-geo (Team:Analytics)

[craigtaverner]

There are two solutions to this:

• Use the new geometry simplifier library added at Geometry simplifier (streaming) #94859
• Use the JTS simplifier used in the MVT API

The second option is a better option for a few reasons:

• JTS is already geared towards simplifying full-geometries, is suitable for simplifications based on resolution/level of detail, and is a well tried and tested solution
• The new algorithm is focused on streaming data, only limits to total point count (not resolution or level of detail) and is a new algorithm, not well tried and tested.
• JTS is used in MVT, and so is already use for similar kibana visualizations