Add map/reduce property aggregation (Supercluster parity)

[StoneTapeStudios]

Supercluster supports map and reduce options that let users aggregate custom properties per cluster. For example, summing a numeric property across all points in a cluster:

const index = new Supercluster({
  map: (props) => ({ sum: props.myValue }),
  reduce: (accumulated, props) => {
    accumulated.sum += props.sum;
  },
});

This is one of Supercluster's most-used features. arrow-supercluster doesn't support it yet.

What needs to happen

• Add optional map and reduce fields to ArrowClusterEngineOptions in packages/arrow-supercluster/src/types.ts
• During clustering in _cluster() (packages/arrow-supercluster/src/arrow-cluster-engine.ts), call map for leaf points and reduce when merging points into clusters
• Store the aggregated properties and expose them in ClusterOutput (likely a new properties array or similar)
• The tricky part: ClusterOutput currently uses only typed arrays. Aggregated properties are user-defined objects, so this needs a design decision on how to expose them without breaking the typed-array-only contract
• Add tests in packages/arrow-supercluster/tests/

References

• Supercluster map/reduce docs: https://github.com/mapbox/supercluster#property-mapreduce-options
• Current engine implementation: packages/arrow-supercluster/src/arrow-cluster-engine.ts (the _cluster method at ~line 354)