Remove ordinal grouping operator (#131133)

The ordinals grouping operator was introduced to speed up aggregation 
before ordinal blocks and related optimizations in block hashes were
available. However, this operator has several issues:

1. It only supports single grouping with the `keyword` type and requires 
`doc_values`.

2. It needs a separate aggregation implementation, which currently lacks 
test coverage. We had performance issues with the `VALUES` aggregation
using this operator (see #130576).

3. It can be slower and use more memory when the target documents have 
sparse cardinality (see #98963).

4. Ad-hoc planning, although this can now be addressed with local plans.

Although the ordinals grouping operator is slightly faster than the hash 
operator with ordinal blocks, its complexity now outweighs the benefits.
This PR proposes removing the operator. Below is the NYC_taxis
benchmark.

Closes #98963

Author: dnhatn

benchmarks/src/main/java/org/elasticsearch/benchmark/compute/operator/ValuesAggregatorBenchmark.java

@@ -95,8 +95,7 @@ static void selfTest() {
         try {
             for (String groups : ValuesAggregatorBenchmark.class.getField("groups").getAnnotationsByType(Param.class)[0].value()) {
                 for (String dataType : ValuesAggregatorBenchmark.class.getField("dataType").getAnnotationsByType(Param.class)[0].value()) {
-                    run(Integer.parseInt(groups), dataType, 10, 0);
-                    run(Integer.parseInt(groups), dataType, 10, 1);
+                    run(Integer.parseInt(groups), dataType, 10);
                 }
             }
         } catch (NoSuchFieldException e) {
@@ -114,10 +113,7 @@ static void selfTest() {
     @Param({ BYTES_REF, INT, LONG })
     public String dataType;
 
-    @Param({ "0", "1" })
-    public int numOrdinalMerges;
-
-    private static Operator operator(DriverContext driverContext, int groups, String dataType, int numOrdinalMerges) {
+    private static Operator operator(DriverContext driverContext, int groups, String dataType) {
         if (groups == 1) {
             return new AggregationOperator(
                 List.of(supplier(dataType).aggregatorFactory(AggregatorMode.SINGLE, List.of(0)).apply(driverContext)),
@@ -132,20 +128,8 @@ private static Operator operator(DriverContext driverContext, int groups, String
         ) {
             @Override
             public Page getOutput() {
-                mergeOrdinal();
                 return super.getOutput();
             }
-
-            // simulate OrdinalsGroupingOperator
-            void mergeOrdinal() {
-                var merged = supplier(dataType).groupingAggregatorFactory(AggregatorMode.SINGLE, List.of(1)).apply(driverContext);
-                for (int i = 0; i < numOrdinalMerges; i++) {
-                    for (int p = 0; p < groups; p++) {
-                        merged.addIntermediateRow(p, aggregators.getFirst(), p);
-                    }
-                }
-                aggregators.set(0, merged);
-            }
         };
     }
 
@@ -352,12 +336,12 @@ private static Block groupingBlock(int groups) {
 
     @Benchmark
     public void run() {
-        run(groups, dataType, OP_COUNT, numOrdinalMerges);
+        run(groups, dataType, OP_COUNT);
     }
 
-    private static void run(int groups, String dataType, int opCount, int numOrdinalMerges) {
+    private static void run(int groups, String dataType, int opCount) {
         DriverContext driverContext = driverContext();
-        try (Operator operator = operator(driverContext, groups, dataType, numOrdinalMerges)) {
+        try (Operator operator = operator(driverContext, groups, dataType)) {
             Page page = page(groups, dataType);
             for (int i = 0; i < opCount; i++) {
                 operator.addInput(page.shallowCopy());

x-pack/plugin/esql/compute/src/main/java/org/elasticsearch/compute/aggregation/GroupingAggregator.java

@@ -70,13 +70,6 @@ public void close() {}
         }
     }
 
-    /**
-     * Add the position-th row from the intermediate output of the given aggregator to this aggregator at the groupId position
-     */
-    public void addIntermediateRow(int groupId, GroupingAggregator input, int position) {
-        aggregatorFunction.addIntermediateRowInput(groupId, input.aggregatorFunction, position);
-    }
-
     /**
      * Build the results for this aggregation.
      * @param selected the groupIds that have been selected to be included in

x-pack/plugin/esql/compute/src/main/java/org/elasticsearch/compute/aggregation/GroupingAggregatorFunction.java

@@ -130,6 +130,7 @@ default void add(int positionOffset, IntBlock groupIds) {
 
     /**
      * Add the position-th row from the intermediate output of the given aggregator function to the groupId
+     * TODO: Remove this method as the grouping operator has been removed
      */
     void addIntermediateRowInput(int groupId, GroupingAggregatorFunction input, int position);