Scale gracefully to very large label sets

## Description

Graph Explorer degrades — sometimes severely — when a database has a very large number of distinct node/edge labels (10k+, occasionally 70k+). Schema sync becomes slow or fails, and label-scale UI (selects, schema view) becomes unresponsive. Very large unique-label sets are an uncommon data-modeling anti-pattern, but we should both survive them where the cost is low and inform users when we can't.

## High-Level Plan

Two thrusts under this Initiative:

1. **Scale the mechanics** — more efficient sampling/batching in schema sync, virtualized label-scale UI.
2. **Inform when limits remain** — nudge the user when the combined unique label count is excessive, so they proceed with informed consent.

## Related Issues

- Related to #2076
- Related to #2077
- Related to #2086

---

> [!IMPORTANT]
> Internal only — this issue is maintained by the core team and is not accepting external contributions.

The issue covers Graph Explorer schema sync and label-scale UI, but names no files, tests, or entry points. Start by locating the schema-sync path and the label selectors or schema view, then determine how the related issues divide the work. Done would mean the agreed large-label behavior is implemented and limits are communicated to users.

Written by the indexing model from the issue text.