ESQL: Run aggregations in SINGLE mode when possible

[nik9000]

Description

When you run an aggregation on the coordinating node with ESQL right now you get a plan like:

{
  "operator": "HashAggregationOperator[blockHash=PackedValuesBlockHash{groups=[0:BYTES_REF, 14:BYTES_REF, 15:BYTES_REF, 1:BYTES_REF], entries=1267, size=208.1kb}, aggregators=[GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[2]], mode=INITIAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[3]], mode=INITIAL], GroupingAggregator[aggregatorFunction=MaxLongGroupingAggregatorFunction[channels=[4]], mode=INITIAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[19]], mode=INITIAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[18]], mode=INITIAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[16]], mode=INITIAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[17]], mode=INITIAL]]]",
  "status": {
    "hash_nanos":        7318931,
    "aggregation_nanos": 7684744,
    "pages_processed": 39
  }
},
{
  "operator": "HashAggregationOperator[blockHash=PackedValuesBlockHash{groups=[0:BYTES_REF, 1:BYTES_REF, 2:BYTES_REF, 3:BYTES_REF], entries=1267, size=208.1kb}, aggregators=[GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[4, 5]], mode=FINAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[6, 7]], mode=FINAL], GroupingAggregator[aggregatorFunction=MaxLongGroupingAggregatorFunction[channels=[8, 9]], mode=FINAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[10, 11]], mode=FINAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[12, 13]], mode=FINAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[14, 15]], mode=FINAL], GroupingAggregator[aggregatorFunction=MaxBytesRefGroupingAggregatorFunction[channels=[16, 17]], mode=FINAL]]]",
  "status": {
    "hash_nanos":         838334,
    "aggregation_nanos": 1164512,
    "pages_processed": 1
  }
},

It feels like the second one is just a waste. In this particular query we're only burning 2ms, but this feels like it'd be much bigger for bigger queries.

[elasticsearchmachine]

Pinging @elastic/es-analytical-engine (Team:Analytics)

[dnhatn]

+1. I had #106148 for both aggregation and topN.