Add list_scalar API

[isVoid]

Closes #5887

This PR adds list_scalar, a scalar for a LIST type column. A list_scalar contains a validity field and a _data field. The validity field describe whether the list_scalar is valid as a whole, and is accessible via derived is_valid member function. The _data field is a cudf::column, owned by the scalar class. The data column can be either nested or non-nested type, depending on what LIST column it originate from.

For example, a scalar from a List<int> column contains a int column:

C = List<int> column: {{1, 2, 3}, {4}, {5, 6}}
C[0] = int column: {1, 2, 3}

A scalar from a List<List<int>> column would contain a List<int> column instead:

C = List<List<int>> column: {{{1, 2}, {}}, {{3}}, {{4, 5}, {6}}}
C[2] = List<int> column: {{4, 5}, {6}}

list_scalar has a view method, which returns a immutable view to the underlying data.

[codecov]

Codecov Report

Merging #7584 (eaa8024) into branch-0.20 (51336df) will increase coverage by 0.02%.
The diff coverage is 87.91%.

@@               Coverage Diff               @@
##           branch-0.20    #7584      +/-   ##
===============================================
+ Coverage        82.88%   82.91%   +0.02%     
===============================================
  Files              103      103              
  Lines            17668    17658      -10     
===============================================
- Hits             14645    14642       -3     
+ Misses            3023     3016       -7     

Impacted Files	Coverage Δ
python/cudf/cudf/core/column/__init__.py	100.00% <ø> (ø)
python/cudf/cudf/utils/cudautils.py	57.75% <25.00%> (ø)
python/cudf/cudf/core/column/column.py	88.64% <71.42%> (ø)
python/cudf/cudf/core/column/numerical.py	94.43% <72.72%> (ø)
python/cudf/cudf/core/dataframe.py	90.87% <83.33%> (+0.01%)	⬆️
python/cudf/cudf/utils/utils.py	89.53% <91.66%> (+0.02%)	⬆️
python/cudf/cudf/core/column/datetime.py	89.91% <100.00%> (ø)
python/cudf/cudf/core/column/timedelta.py	88.66% <100.00%> (ø)
python/cudf/cudf/core/groupby/groupby.py	92.28% <100.00%> (+0.83%)	⬆️
python/cudf/cudf/core/index.py	93.07% <100.00%> (ø)
... and 14 more

---

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 b8baed5...eaa8024. Read the comment docs.

[harrism]

This new class needs to be tested...

[devavret]

Removed my comments because they were a subset of @harrism 's review comments.

1. No need for two separate view() and value() methods. This will probably not return a host usable value like the rest of the scalar classes. Any such member can be added later when needed.
2. I'd vote against having validity as an enum.

[devavret]

Is this meant to support a list of list case? If so, you might want to use list_device_view. @mythrocks would be able to help explain.

[isVoid]

@devavret yes this should support nested lists case. Would love to have more clarification of list_view and list_device_view.

[isVoid]

Addressing @devavret and @jrhemstad 's comment, this PR will not introduce an is_valid enum class. Addressing @harrism 's comment, value and view is combined into a single view function, that returns a list_view. Would like to have clarification for:

1. Should a function that returns list_device_view be added?
2. For testing the scalar data equality - I would propose a _data field accessor for list_view that returns a column_view, and resort to expect_columns_equal to check for data equality.

[devavret]

1. Should a function that returns list_device_view be added?

After an offline discussion, we figured we should return a column_view for now and hopefully revisit when list_view is more fleshed out.

[harrism]

After an offline discussion, we figured we should return a column_view for now and hopefully revisit when list_view is more fleshed out.

We should be careful and not rush this. I don't understand all the details, but rushing this could result in painful churn, so let's figure out what the right thing to do is. If figuring that out is too big a task, then we can consider a halfway solution.

[harrism]

@gpucibot merge