[VL] Add Spark 4.1 TimeType support

[malinjawi]

What changes are proposed in this pull request?
This PR adds native Spark 4.1 TIME data type support in the Velox backend by mapping Substrait TIME to Velox TIME_MICRO_UTC and handling TIME literals end to end.

The change:

• adds Gluten Substrait TimeType and TimeLiteral nodes
• converts Spark 4.1 TimeType literals from Spark nanos to Substrait/Velox micros
• parses Substrait Type.Time as Velox TIME_MICRO_UTC in the C++ Substrait parser
• maps TIME_MICRO_UTC through Velox/Substrait signatures and Velox-to-Substrait type conversion
• converts Velox TIME_MICRO_UTC values back to Spark UnsafeRow nanos in columnar-to-row, including nested arrays, maps, and rows
• allows supported TIME precision through Velox validation and planning paths
• adds Java, Scala, and C++ coverage for time literals, type conversion, signatures, and columnar-to-row serialization

Why are the changes needed?
#11919 tracks Spark 4.1 TimeType support for the Velox backend. Spark 4.1 adds TimeType to DataTypeTestUtils.atomicTypes, so tests and planning paths that enumerate atomic types can hit TIME even when Gluten/Velox cannot yet parse or execute it. Mapping Substrait TIME to Velox TIME_MICRO_UTC gives Gluten a native representation for supported Spark TIME precision and unblocks literal/planning paths.

Does this PR introduce any user-facing change?
No public API change. It extends native Spark 4.1 TIME support in the Velox backend.

How was this patch tested?
Built and ran locally:

• Spark 4.1 Velox compile with Scala 2.13 and Java 21
• focused Spark 4.1 VeloxLiteralSuite Time Literal test
• C++ Velox backend build
• generic_benchmark smoke run

Commands:

• git diff --cached --check
• mvn -Pbackends-velox,spark-4.1,scala-2.13,java-21 -pl backends-velox -am -DskipTests -Dcheckstyle.skip -Dspotless.check.skip compile
• ./dev/run-scala-test.sh --force -Pjava-21,spark-4.1,scala-2.13,backends-velox,hadoop-3.4,spark-ut,delta -pl backends-velox -s org.apache.gluten.execution.VeloxLiteralSuite -t "Time Literal" -Dscalastyle.skip=true -Dcheckstyle.skip -Dspotless.check.skip
• cmake --build cpp/build --target velox -j 4

Native C++ unit-test targets were not run locally because this checkout does not have the Velox vector test utility archive built (libvelox_vector_test_lib.a).

Performance
I ran targeted local benchmarks comparing clean upstream main against this patch.

Focused columnar-to-row benchmark:

type	before median us	after median us	delta
BIGINT	1325.407	1329.853	+0.335%
TIME_MICRO_UTC	1331.705	1643.357	+23.403%

The TIME_MICRO_UTC delta is expected because this patch adds the required micros-to-nanos normalization for Spark UnsafeRow compatibility; the baseline treated TIME as a raw BIGINT-like value.

Broad generic_q5 smoke benchmark:

metric	before	after	delta
real median ms	29.076	29.159	+0.285%
CPU median ms	28.993	28.991	-0.006%

Related issue: #11919
Tracked by #11910

Was this patch authored or co-authored using generative AI tooling?
Generated-by: IBM BOB

[github-actions]

Run Gluten Clickhouse CI on x86

[github-actions]

Run Gluten Clickhouse CI on x86