Support null_policy::EXCLUDE for COLLECT rolling aggregation

[mythrocks]

Closes #7258.

#7189 implements COLLECT aggregations to be done from window functions. The semantics of how null input rows are handled are consistent with CUDF semantics.
E.g.

auto input_col = fixed_width_column_wrapper<int32_t>{70, ∅, 72, 73, 74};
auto output_col = cudf::rolling_window(input_col, 2, 1, 1, collect_aggr);
            // == [ [70,∅], [70,∅,72], [∅,72,73], [72,73,74], [73,74] ]

Note that the null element (∅) is replicated in the first 3 rows of the output.

SparkSQL (and Hive, and other big data SQL systems) have different semantics, in that all null elements are purged. The output for the same operation should yield the following:

auto sparkish_output_col = cudf::rolling_window(input_col, 2, 1, 1, collect_aggr);
            // == [ [70], [70,72], [72,73], [72,73,74], [73,74] ]

CUDF should allow the COLLECT aggregation to be constructed with an optional null_policy argument (with default INCLUDE). The COLLECT window function should check the policy, and filter out null list-elements a posteriori.

[mythrocks]

The build failure might be because branch-0.19 is quite new:

22:41:05 PackagesNotFoundError: The following packages are not available from current channels:
22:41:05 
22:41:05   - ucx-py=0.19
22:41:05   - rapids-build-env=0.19
22:41:05   - rapids-notebook-env=0.19
22:41:05 
22:41:05 Current channels:...

[mythrocks]

I just checked that there isn't a pressing need to similarly exclude NaN values from a collected list, at least from SparkSQL. Spark retains NaN values.

[mythrocks]

rerun tests

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (branch-0.19@846e6b6). Click here to learn what that means.
The diff coverage is n/a.

@@              Coverage Diff               @@
##             branch-0.19    #7264   +/-   ##
==============================================
  Coverage               ?   82.19%           
==============================================
  Files                  ?      100           
  Lines                  ?    16968           
  Branches               ?        0           
==============================================
  Hits                   ?    13947           
  Misses                 ?     3021           
  Partials               ?        0           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 846e6b6...60aa7c6. Read the comment docs.

[rgsl888prabhu]

Jake has almost covered all essentials, have only minor question.

[vuule]

Just a few suggestions, already looks great!

[mythrocks]

@gpucibot merge

[mythrocks]

Thanks for the reviews, all. I've learnt a bit on this one.