Fix quantile overflow: wrap p99 query with toInt64() for JDBC compat

quantile(0.99) returns Float64 which ClickHouse JDBC cannot cast to
Long directly. Same toInt64() pattern already used in timeseries query.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

cameleer3-server-app/src/main/java/com/cameleer3/server/app/search/ClickHouseSearchEngine.java

@@ -90,7 +90,7 @@ public class ClickHouseSearchEngine implements SearchEngine {
     @Override
     public ExecutionStats stats(Instant from, Instant to) {
         Long p99 = jdbcTemplate.queryForObject(
-                "SELECT quantile(0.99)(duration_ms) FROM route_executions " +
+                "SELECT toInt64(quantile(0.99)(duration_ms)) FROM route_executions " +
                 "WHERE start_time >= ? AND start_time <= ?",
                 Long.class, Timestamp.from(from), Timestamp.from(to));
         Long active = jdbcTemplate.queryForObject(