feat(query): push rank limit into aggregate partial node

[sundy-li]

I hereby agree to the terms of the CLA available at: https://docs.databend.com/dev/policies/cla/

Summary

This pr pushes rank limit into aggregate partial node.

For example:

SELECT UserID, SearchPhrase, COUNT(*) FROM hits GROUP BY UserID, SearchPhrase ORDER BY UserID,  SearchPhrase 
 LIMIT 10;

As the group keys may have high cardinality, the HT will resize multiple times to hold the coming keys which is inefficient.

Now we will filter group keys using partial sort with LimitType::Rank(n) to keep top n unique keys.

So we inject a SortPartialTransform before partial agg now.

🐳 :) explain pipeline SELECT UserID, SearchPhrase, COUNT(*) FROM hits GROUP BY UserID, SearchPhrase ORDER BY UserID,  SearchPhrase
 LIMIT 10;
-[ EXPLAIN ]-----------------------------------
CompoundBlockOperator(Project) × 1
  LimitTransform × 1
    Merge to MultiSortMerge × 1
      TransformSortSpill × 16
        TransformSortMergeLimit × 16
          SortPartialTransform × 16
            TransformFinalAggregate × 16
              TransformSpillReader × 16
                Merge to Resize × 16
                  Merge to TransformPartitionBucket × 1
                    TransformAggregateSpillWriter × 16
                      TransformPartialAggregate × 16
                        SortPartialTransform × 16
                          NativeDeserializeDataTransform × 16
                            SyncReadNativeDataSource × 16

Performance in my PC is 2.8 sec --> 0.35 sec.

Tests

• [x] Unit Test
• [x] Logic Test
• [ ] Benchmark Test
• [ ] No Test - Explain why

Type of change

• [ ] Bug Fix (non-breaking change which fixes an issue)
• [ ] New Feature (non-breaking change which adds functionality)
• [ ] Breaking Change (fix or feature that could cause existing functionality not to work as expected)
• [ ] Documentation Update
• [ ] Refactoring
• [x] Performance Improvement
• [ ] Other (please describe):

---

This change is