[FEA][Java][JNI] Support multi-input groupby aggregations

[malinjawi]

Description

Adds role-tagged Java/JNI groupby aggregations that allow a logical
multi-input aggregate to be represented by multiple Aggregation instances
sharing a correlation id. The JNI groupby path buckets those role-tagged
aggregations, dispatches the underlying libcudf request directly, and places
results back in the requested output slots.

This PR adds the initial min_by roles:

• GroupByAggregation.orderingForMinBy(long multiInputId)
• GroupByAggregation.valueForMinBy(long multiInputId)
• MultiInputIds.next() for thread-safe id generation
• JNI-native multi_input_aggregation dispatch for paired min-by inputs

This avoids forcing callers to pack multi-input aggregate arguments into a
struct column before groupby and unpack the struct afterward.

closes #22276

Validation

• clang-format -i on touched native C++ files.
• git diff --check rapids/main...HEAD.
• Java main source compilation with javac -source 8 -target 8.
• Focused Java test compilation for MultiInputAggregationTest.
• mvn -B surefire:test -Dtest="ai.rapids.cudf.MultiInputAggregationTest#testMultiInputIdsUniqueness+testMinByAggregationEquality" -DfailIfNoTests=false in the cuDF Java devcontainer.
• mvn -B install -DskipTests ... in the cuDF Java devcontainer, including native cudfjni rebuild of AggregationJni.cpp and TableJni.cpp.

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[malinjawi]

I do not have permission to add RAPIDS repo labels from this fork. Could a maintainer please add the required label-checker labels? Suggested labels: feature request, Java, Spark, and non-breaking.