…uild

- Change doc_key function to return id directly instead of prefixed format
- Update rebuild_meta_index to filter out reserved keys instead of using prefix matching
- Add reserved keys array ["meta", "_graph"] to exclude from document processing

- Add reasoning_index field to IndexedDocument struct to store
  pre-computed reasoning index for retrieval acceleration
- Initialize reasoning_index as None in IndexedDocument default
- Pass reasoning_index from indexing result to document and persist it

perf(summary): improve token estimation and adjust shortcut threshold

- Add token generation tracking when summaries are created
- Reduce shortcut_threshold from 200 to 50 to enable faster processing
  for smaller documents

…tform architecture

Complete rewrite of the example content to provide comprehensive documentation
for the distributed data processing platform architecture. The new content
covers system architecture, ingestion layer, processing engine, storage
layer with tiering mechanisms, and query serving layer. Also includes
deployment and operations details with monitoring and tracing setup.

- Add comprehensive documentation for the distributed data processing
  platform architecture
- Include details about layered system architecture with clear separation
  of concerns between ingestion, processing, storage, and serving layers
- Document microservices-based approach with independent scaling
  capabilities for cost-effective resource utilization
- Cover real-time streaming and batch processing paradigms support
- Add information about deployment and operations on Kubernetes
- Include monitoring and observability details with Prometheus,
  Grafana, and OpenTelemetry