[ES|QL] Refactor MV_UNION and MV_INTERSECTION to use shared set operation helper

[mridula-s109]

Follow-up to #139664 (MV_UNION implementation).

This PR extracts common set operation logic into a shared MvSetOperationHelper utility class to reduce code duplication between MV_UNION and MV_INTERSECTION.

Changes

• Created MvSetOperationHelper with union() and intersection() static methods
• Refactored MvUnion.processUnionSet() andMvIntersection.processIntersectionSet() to use the helper

[mridula-s109]

@markjhoy and @ioanatia I have verified by running all the tests locally as well. I haven't added unit tests for the helper explicitly as we have sufficient tests for union and intersection functionality. Please do have a look and let me know!

[elasticsearchmachine]

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

[markjhoy]

Thanks for this @mridula-s109 -- this looks clean. The only thing I think we can optimize here is to not have to initialize additional sets in the helper class... I'm wondering if something like creating a Set<E> class that extends the LinkedHashSet, and assigning the first set of values to it would be better.... I'm thinking something like:

    public class OperationalSet<E> extends LinkedHashSet<E> {
        public Set<E> union(Set<E> set) {
            this.addAll(set);
            return this;
        }
        
        public Set<E> intersect(Set<E> set) {
            this.retainAll(set);
            return this;
        }
    }

Set<T> firstSet = new OperationalSet<>();
for (int i = 0; i < firstValueCount; i++) {
    values.add(getValueFunction.apply(firstValueIndex + i, field1));
}

Set<T> secondSet = new LinikedHashSet<>();
int secondValueIndex = field2.getFirstValueIndex(position);
for (int i = 0; i < secondValueCount; i++) {
    secondSet.add(getValueFunction.apply(secondValueIndex + i, field2));
}

Set<T> result = firstSet.union(secondSet);
(or - Set<T> result = firstSet.intersection(secondSet); )

[mridula-s109]

public class OperationalSet extends LinkedHashSet {
public Set union(Set set) {
this.addAll(set);
return this;
}

    public Set<E> intersect(Set<E> set) {
        this.retainAll(set);
        return this;
    }
}

Set firstSet = new OperationalSet<>();
for (int i = 0; i < firstValueCount; i++) {
values.add(getValueFunction.apply(firstValueIndex + i, field1));
}

Set secondSet = new LinikedHashSet<>();
int secondValueIndex = field2.getFirstValueIndex(position);
for (int i = 0; i < secondValueCount; i++) {
secondSet.add(getValueFunction.apply(secondValueIndex + i, field2));
}

Set result = firstSet.union(secondSet);
(or - Set result = firstSet.intersection(secondSet); )

thanks @markjhoy for the suggestion. i have addressed the code optimisation.

[markjhoy]

Nice. In the words of Marie Kondo - "This sparks joy" 😂