feat(console): add concurrency cap or pagination for cross-catalog search data fetching

##Context

The search feature https://github.com/apache/polaris-tools/pull/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

Start with the console search dialog and the client-side fetch path described in PR #187, tracing catalog, namespace, table, and view requests. Decide whether the issue's concurrency-cap or pagination approach fits the existing search flow, then verify that opening search avoids the current request spike while still loading the required results.

Written by the indexing model from the issue text.