Add `statistics_by_partition` API to `ExecutionPlan`

[xudong963]

After #15432 is merged, we'll have partition-level statistics in DataSource.

To make it be used by query optimization, we should flow the info to other nodes, such as SPM, union, etc, so we can add the statistics_by_partition API to ExecutionPlan to implement the idea.

[alamb]

For anyone else following along, this is part of work to optimize ORDER by queries with pre-sorted data

• [EPIC] Avoid sort for already sorted Parquet files that do not overlap values on condition #6672

Specifically I believe it is a pre-requisite for determining non overlapping key ranges;

• Analysis to supportSortPreservingMerge --> ProgressiveEval #15191