[Maps] Introduce fields

[thomasneirynck]

WIP

Do not merge. This is for exploration only. Work will be split up into multiple smaller PRs.

Prereq:

• [Maps] Introduce Agg Source #49782
• [Maps] Introduce typed styles #49803

---

This makes following changes to the code base.

Introduces field interface

  getName() //unique name of the field
  isValid() //whether it is usable (e.g. partial field configs can arise when editing)
  async getType()  //data-type of the field, string, number, etc...
  async getLabel() //display of the field
  async createTooltipProperty()//create corresponding tooltip entry

Introduces 4 field types (x-pack/legacy/plugins/maps/public/layers/fields)

• EMSRegionField: fields from EMS
• ESAggField: fields that wrap results of aggs (e.g. geogrid-sources, or term-join metrics)
• ESDocField: fields from ES-index pattern
• KibanaRegionField: todo

Introduces AbstractESAggSource class (DONE in #49782)

• top-level for ESGeoGridSource, ESPewPewSource, ESTermSource
• Removes agg-formatting duplication and field-creation

Organizes style-modules and related react-components by type (DONE in #49803)

• vector style modules: x-pack/legacy/plugins/maps/public/layers/styles/heatmap
• heatmap style modules: x-pack/legacy/plugins/maps/public/layers/styles/vector

Add typing for all vector-style-properties (done in #49803)

(x-pack/legacy/plugins/maps/public/layers/styles/vector/properties)

• dynamic_*_property: style-properties bound to a field
• static_*_property: static properties

This also removes the duplicated branching-code from all mb-syncing operations

---

Introduce fields as top-level concept.

Fields are scattered and inconsistent throughout the code-base right now, wrt. display, config, and typing.

Fields should be used consistently for:

• tooltips
• top hits sort and split field editor
• joins
  -- left field definitions
  -- right field definitions
• dynamic styling properties
• agg metrics
  -- for grid source
  -- for pewpew source

Fields are source-dependent

• EMS
• ES fields (agg results and docs)
• Kibana layer configs

Later, fields should also be used for

• automagic suggestion of styling params (e.g. categorical vs ordinal styling)
• preloading of data-ranges (e.g. to determine intelligent breaks in color ramp)
• identifying whether field is id-field or display-field (this is EMS top-level metadata, unused by Maps)

[elasticmachine]

Pinging @elastic/kibana-gis (Team:Geo)

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request
• Commit: 659280e

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request
• Commit: 259f727

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request
• Commit: ced272e

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request
• Commit: 64ceb0d

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request
• Commit: 3cbe940

[elasticmachine]

💚 Build Succeeded

• continuous-integration/kibana-ci/pull-request
• Commit: ab9bd8e

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request
• Commit: a70ccff

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request
• Commit: 6e0e168

[monfera]

This seems to relate to discussions in #35481 and #44955 and plans to think up the spiritual successor of index patterns (about which there'll be a session next week)

[nreese]

This seems to relate to discussions in #35481 and #44955 and plans to think up the spiritual successor of index patterns (about which there'll be a session next week)

No, this is unrelated to those discussions. This PR just adds a fields class to the Maps app to consolidate custom labels, formatters, and meta data like max and min and percentiles.

The Maps app still uses Index patterns heavily.

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request
• Commit: 066865d

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request
• Commit: d4b1808

[thomasneirynck]

This was getting split up into multiple PRs. Closing in favor of #50044