[SPARK-42236][SQL] Refine NULLABLE_ARRAY_OR_MAP_ELEMENT

[itholic]

What changes were proposed in this pull request?

This PR proposes to refine NULLABLE_ARRAY_OR_MAP_ELEMENT into main-sub classes structure.

NOT_NULL_CONSTRAINT_VIOLATION

• ARRAY_ELEMENT
• MAP_VALUE

Why are the changes needed?

The name of error class is misleading, and we can make this more generic so that we reuse for various situation.

Does this PR introduce any user-facing change?

No.

How was this patch tested?

Updated & added UTs.

[itholic]

We could also add NOT_NULL_CONSTRAINT_VIOLATION.STRUCT_FIELD for non-nullable StructField, but this case now it's buried under DATATYPE_MISMATCH.CAST_WITHOUT_SUGGESTION as below:

spark-sql> SELECT CAST(named_struct('a', null) AS STRUCT<a int not null>);
[DATATYPE_MISMATCH.CAST_WITHOUT_SUGGESTION] Cannot resolve "named_struct(a, NULL)" due to data type mismatch: cannot cast "STRUCT<a: VOID>" to "STRUCT<a: INT>".; line 1 pos 7;
'Project [unresolvedalias(cast(named_struct(a, null) as struct<a:int>), None)]

Let me migrate this case in separate ticket to facilitate review since the change would requires a non-trivial amount of code modifications.

[itholic]

cc @MaxGekk @srielau FYI

[MaxGekk]

@itholic Could you fix, please:

org.scalatest.exceptions.DuplicateTestNameException: Duplicate test name: DATATYPE_MISMATCH.INVALID_JSON_SCHEMA: invalid top type passed to from_json()

[MaxGekk]

+1, LGTM. Merging to master/3.4.
Thank you, @itholic.