Avoid aggregate repartitioning for audited date_trunc on single-key timestamp ranges #26448

Grouped aggregation can avoid a hash shuffle when transforming a range-partitioned timestamp keeps every group within one input partition.

For example, suppose a table is range-partitioned on timestamp, with partitions split at hourly boundaries:

SELECT
    key,
    date_trunc('hour', timestamp) AS hour,
    SUM(value)
FROM events
GROUP BY key, date_trunc('hour', timestamp);

Each (key, hour) group belongs entirely to one partition, so aggregation can finish independently in each partition. Changing the grouping expression to date_trunc('day', timestamp) would combine rows from multiple hourly partitions and still require a shuffle.

Initially support audited, timezone-less date_trunc expressions over single-key timestamp ranges. Require an explicit, default-deny function capability and validate concrete split boundaries before removing repartitioning. Unsafe or unknown cases must retain the shuffle, and the aggregate-locality proof must not establish join-layout compatibility.

Part of #23569 and the staged work in #25344 and #25345.