[VL] Enable array_repeat & array_except function

[PHILO-HE]

What changes were proposed in this pull request?

These functions should have been supported: array_repeat, array_except, arrray_distinct, array_position. Two of them only need few code to get them correctly mapped to velox function. Also added tests to verify they are offoaded.

How was this patch tested?

Spark UT & new test.

[github-actions]

Thanks for opening a pull request!

Could you open an issue for this pull request on Github Issues?

https://github.com/oap-project/gluten/issues

Then could you also rename commit message and pull request title in the following format?

[GLUTEN-${ISSUES_ID}][COMPONENT]feat/fix: ${detailed message}

See also:

• Other pull requests

[github-actions]

Run Gluten Clickhouse CI

[github-actions]

Run Gluten Clickhouse CI

[github-actions]

Run Gluten Clickhouse CI

[github-actions]

Run Gluten Clickhouse CI

[github-actions]

Run Gluten Clickhouse CI

[github-actions]

Run Gluten Clickhouse CI

[PHILO-HE]

A velox PR is fixing the test issue reported for array_repeat: facebookincubator/velox#8630.

�[31m- SPARK-36753: ArrayExcept should handle duplicated Double.NaN and Float.Nan *** FAILED ***�[0m
2024-01-24T07:34:28.7280340Z �[31m  Incorrect evaluation: array_except([NaN,1.0], [NaN]), actual: WrappedArray(NaN, 1.0), expected: List(1.0) (GlutenTestsTrait.scala:288)�[0m

�[31m- ArrayRepeat *** FAILED ***�[0m
2024-01-24T07:34:19.3744283Z �[31m  org.apache.spark.SparkException: Job aborted due to stage failure: Task 1 in stage 347.0 failed 1 times, most recent failure: Lost task 1.0 in stage 347.0 (TID 695) (8c96ae9a310d executor driver): io.glutenproject.exception.GlutenException: java.lang.RuntimeException: Exception: VeloxUserError�[0m
2024-01-24T07:34:19.3746335Z �[31mError Source: USER�[0m
2024-01-24T07:34:19.3746817Z �[31mError Code: INVALID_ARGUMENT�[0m
2024-01-24T07:34:19.3747669Z �[31mReason: (-1 vs. 0) Count argument of repeat function must be greater than or equal to 0�[0m
2024-01-24T07:34:19.3748471Z �[31mRetriable: False�[0m
2024-01-24T07:34:19.3748917Z �[31mExpression: count >= 0�[0m
2024-01-24T07:34:19.3749508Z �[31mContext: repeat(hi:VARCHAR, -1:INTEGER)�[0m
2024-01-24T07:34:19.3750128Z �[31mTop-Level Context: Same as context.�[0m
2024-01-24T07:34:19.3750667Z �[31mFunction: checkCount�[0m
2024-01-24T07:34:19.3751222Z �[31mFile: ../../velox/functions/prestosql/Repeat.cpp�[0m
2024-01-24T07:34:19.3751796Z �[31mLine: 52�[0m
2024-01-24T07:34:19.3752264Z �[31mStack trace:�[0m

[github-actions]

Run Gluten Clickhouse CI

[github-actions]

Run Gluten Clickhouse CI

[github-actions]

Run Gluten Clickhouse CI

[github-actions]

Run Gluten Clickhouse CI

[PHILO-HE]

cc @taiyang-li, I directly changed this for CH backend to let these two unsupported functions fall back.

[GlutenPerfBot]

===== Performance report for TPCH SF2000 with Velox backend, for reference only ====

query	log/native_4504_time.csv	log/native_master_02_19_2024_72960de81_time.csv	difference	percentage
q1	34.49	33.43	-1.054	96.94%
q2	24.41	24.73	0.322	101.32%
q3	37.26	37.94	0.684	101.84%
q4	37.21	37.71	0.501	101.35%
q5	70.14	70.80	0.659	100.94%
q6	7.38	7.31	-0.074	98.99%
q7	84.36	82.80	-1.559	98.15%
q8	84.84	86.79	1.956	102.31%
q9	124.07	121.45	-2.624	97.88%
q10	42.49	41.30	-1.196	97.19%
q11	20.55	20.02	-0.526	97.44%
q12	28.40	26.06	-2.338	91.77%
q13	45.05	45.68	0.636	101.41%
q14	20.10	15.10	-5.007	75.09%
q15	27.51	27.77	0.259	100.94%
q16	13.48	14.10	0.611	104.53%
q17	102.98	102.85	-0.125	99.88%
q18	149.02	149.16	0.134	100.09%
q19	14.04	13.92	-0.114	99.19%
q20	26.87	26.87	-0.001	99.99%
q21	223.05	226.51	3.455	101.55%
q22	13.96	13.67	-0.290	97.92%
total	1231.65	1225.96	-5.692	99.54%