[FEA] Fully support nested types in Spark SQL functions

[ttnghia]

Since libcudf now has full support for nested types in both equality and lexicographic comparisons, we should also adopt such features to various Spark SQL functions in the plugin.

Currently, there are several existing FEA issues related to nested types that can be addressed:

Feature	Issue(s)	PR(s)	Status	Notes
Fully support nested type for HashPartitioning	#8676		❌	Top dependency for all other issues.
Verify that array nulls are empty	#5430	#8517	✔️	Top dependency for all other issues.
Order-by on arbitrary level nested arrays	#5509, #7230	#7233	❌	Depends on #5430.
Use merge sort for nested types	#2252	rapidsai/cudf#13347	❌	Depends on #5430 and cudf issues: rapidsai/cudf#8050 and rapidsai/cudf#13514.
Sort arrays containing nested types	#3715		❌	Depends on #5430.
Support hash aggregate on single-level arrays	#6680	#7465	✔️
Support hash aggregate on nested arrays	N/A		❌	Depends on #8676.
Support min/max in groupby/reduction for nested structs	#3153, rapidsai/cudf#8974, rapidsai/cudf#8964	#8638	✔️	There is no support for windowing ops.
Support min/max in groupby/reduction for nested arrays	#4929, #4900, #8668	rapidsai/cudf#13069, rapidsai/cudf#13676, #8689	❌	There is no support for windowing ops.
Support Delta Lake optimized write on Databricks	#7799		❌	Require to implement GPU partitioning for nested types (#4887). Also depends on #5430 and #6680.