##Context
The search feature #187 fetches all catalogs, namespaces, tables, and views client-side when the search dialog opens. The fetch strategy fans out to one request per namespace for tables and views:
1 request per catalog → namespaces
1 request per namespace → tables ← fan-out
1 request per namespace → views ← fan-out
In a deployment with many catalogs this results in several parallel HTTP requests on first open.
Problem
- First open can be noticeably slow in large deployments
- Causes a request spike on the Polaris server
Proposed solution
- Implement a concurrency cap or pagination