Additional tests for cudf::distinct using the new row hasher

[vyasr]

#10641 adds new a new row hasher capable of natively handling nested struct and list types. The new hasher was tested by incorporating it into the implementation of cudf::distinct. Now that the functionality for cudf::distinct has been expanded by the use of this new hasher, we should add tests for some new cases, as was pointed out in some discussion threads on that PR:

• We should add tests of tests with sliced input for both structs and lists input columns (Add row hasher with nested column support #10641 (comment))
• We should add tests for simple nested cases like List<Struct<...>> and Struct<List<...>> to ensure that the new nesting behavior really works as expected (Add row hasher with nested column support #10641 (comment))
• We should add at least one test of a more complex nested case (e.g. List<List<Struct<List<...>>>>) to potentially help catch unexpected errors (Add row hasher with nested column support #10641 (comment))

Adding these tests will be helpful to validate the behavior of the new row hasher.

[vyasr]

@ttnghia I've assigned you since you made this request, but feel free to change that if necessary.

[vyasr]

@bdice I've updated the issue to include some of your testing requests and assigned you as well.