Adds a "static" Facets Dive-like visualization to LIT.

This version supports:

* Grouping the dataset into cells in a matrix
* Labels at the top and on the left of the matrix columns and rows
* Graceful resize that maximizes the space allocated to the matrix while keeping all points visible
* Coloring points in the matrix based on their feature value

PiperOrigin-RevId: 488400430

lit_nlp/api/layout.py

@@ -42,6 +42,7 @@ class LitModuleName(dtypes.EnumSerializableAsValues, enum.Enum):
   SimpleDataTableModule = 'simple-data-table-module'
   DatapointEditorModule = 'datapoint-editor-module'
   SimpleDatapointEditorModule = 'simple-datapoint-editor-module'
+  DiveModule = 'dive-module'
   DocumentationModule = 'documentation-module'
   EmbeddingsModule = 'embeddings-module'
   FeatureAttributionModule = 'feature-attribution-module'

lit_nlp/client/lib/types.ts

@@ -123,7 +123,7 @@ export interface FacetedData {
   /** Name to display */
   displayName?: string;
   /** What values were used as filters to get this data */
-  facets?: FacetMap;
+  facets: FacetMap;
 }
 
 /**