Configurable openCypher schema-sync batch size in connection settings

## Description

openCypher schema sync will batch attribute sampling into a fixed number of labels per request (see #2076). The batch size will ship as a hard-coded default (100). On graphs with very large label counts, or against smaller/older Neptune instances, the ideal batch size may differ — too large risks slow requests or timeouts, too small leaves round-trips on the table. Today there is no way to tune it without a code change and redeploy.

## Preferred Solution

Expose the openCypher schema-sync batch size as a per-connection setting in the connection settings UI, defaulting to the shipped value. This lets an operator raise it to cut round-trips on a fast, high-label graph, or lower it if a specific instance struggles, without rebuilding.

Notes:

- Scope to the openCypher connector's schema-sync batching for now; Gremlin has its own batch constant and is out of scope unless we later unify them.
- Live testing on a ~10k-label Neptune 1.4.7.0 graph showed no memory ceiling — the constraint is request time — so the setting is a tuning knob, not a correctness fix.

## Related Issues

- Blocked by #2076
- Related to #1681

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

The issue names the connection settings UI and the openCypher connector's schema-sync batching; start by tracing those entry points and the fixed default described in #2076. Done means the per-connection setting defaults to 100 and lets operators tune openCypher batching without changing Gremlin behavior or rebuilding.

Written by the indexing model from the issue text.