[SPARK-38734][SQL] Remove the error class INDEX_OUT_OF_BOUNDS

[MaxGekk]

What changes were proposed in this pull request?

In the PR, I propose to remove the error class INDEX_OUT_OF_BOUNDS from error-classes.json and the exception SparkIndexOutOfBoundsException. And replace the last one by a SparkException w/ the error class INTERNAL_ERROR because the exception should not be raised in regular cases.

ArrayDataIndexedSeq throws the exception from apply(), and ArrayDataIndexedSeq can be created from ArrayData.toSeq only. The last one is invoked from 2 places:

1. The Slice expression ( or slice function):
   

   spark/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/collectionOperations.scala

   Lines 1600 to 1601 in 443eea9

   	val data = arr.toSeq[AnyRef](elementType)
   	new GenericArrayData(data.slice(startIndex, startIndex + lengthInt))

where any access to the produced array is guarded:

spark-sql> set spark.sql.ansi.enabled=true;
spark.sql.ansi.enabled	true
Time taken: 2.415 seconds, Fetched 1 row(s)
spark-sql> SELECT slice(array(1, 2, 3, 4), 2, 2)[4];
...
org.apache.spark.SparkArrayIndexOutOfBoundsException: [INVALID_ARRAY_INDEX] The index 4 is out of bounds. The array has 2 elements. Use the SQL function `get()` to tolerate accessing element at invalid index and return NULL instead. If necessary set "spark.sql.ansi.enabled" to "false" to bypass this error.
== SQL(line 1, position 8) ==
SELECT slice(array(1, 2, 3, 4), 2, 2)[4]
       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

	at org.apache.spark.sql.errors.QueryExecutionErrors$.invalidArrayIndexError(QueryExecutionErrors.scala:239)
	at org.apache.spark.sql.catalyst.expressions.GetArrayItem.nullSafeEval(complexTypeExtractors.scala:271)

see

spark/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/complexTypeExtractors.scala

Lines 268 to 271 in a9bb924

	if (index >= baseValue.numElements() || index < 0) {
	if (failOnError) {
	throw QueryExecutionErrors.invalidArrayIndexError(
	index, baseValue.numElements, getContextOrNull())

2. MapObjects.convertToSeq:
   

   spark/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/objects/objects.scala

   Line 886 in 5b96e82

   _.asInstanceOf[ArrayData].toSeq[Any](et)

where any access to the produced IndexedSeq is guarded via map-way access in

spark/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/objects/objects.scala

Lines 864 to 867 in 5b96e82

	inputCollection.iterator.map { element =>
	row.update(0, element)
	lambdaFunction.eval(row)
	}

Why are the changes needed?

To improve code maintenance.

Does this PR introduce any user-facing change?

No.

How was this patch tested?

By running the affected test suite:

$ build/sbt "core/testOnly *SparkThrowableSuite"
$ build/sbt "test:testOnly *ArrayDataIndexedSeqSuite"

[MaxGekk]

cc @panbingkun

[MaxGekk]

Merging to master. Thank you, @HyukjinKwon @cloud-fan for review.