Phase 1 of unified storage search sharding implementation:
- Add SubIndexKey type extending NamespacedResource with SubIndexID
- Add new settings: sub_indexes_per_namespace, large_folder_threshold
- Add subIndexCache to bleveBackend for sharded index storage
- Add GetSubIndexForDocument() using FNV32a hash for consistent distribution
- Add GetAllSubIndexKeys() for scatter-gather queries
- Implement compositeIndex wrapper that:
- Routes BulkIndex to correct sub-index based on document key hash
- Uses Bleve IndexAlias for unified search across sub-indexes
- Aggregates DocCount across all sub-indexes
- Delegates UpdateIndex to all sub-indexes
This enables horizontal scaling for namespaces with 1M+ resources by
distributing documents across multiple sub-indexes (recommended: 64).
Rafael Paulovic
aa90ac7ccc