This is in prep for rapidsai/cudf#11968 and
rapidsai/cudf#13372.

`libcudf` will soon require that all CUDF columns are created with a
known null-count. `UNKNOWN_NULL_COUNT` will no longer be supported,
or even available as a code constant.

This change replicates part of rapidsai/cudf#13355,
as it applies to `row_conversion.cu`. The (single) reference to
the unknown-null-count is replaced with a pre-calculated value.

Signed-off-by: MithunR <mythrocks@gmail.com>

This is in prep for rapidsai/cudf#11968 and
rapidsai/cudf#13372.

`libcudf` will soon require that all CUDF columns are created with a
known null-count. `UNKNOWN_NULL_COUNT` will no longer be supported,
or even available as a code constant.

This change replicates part of rapidsai/cudf#13355,
as it applies to `row_conversion.cu`. The (single) reference to
the unknown-null-count is replaced with a pre-calculated value.

Signed-off-by: MithunR <mythrocks@gmail.com>

This is a followup to NVIDIA#1148.

`row_conversion.cu` was modified in rapidsai/cudf#13372
to explicitly calculate null-counts for output columns.

This commit replicates the changes in cudf/pull/13372, and adds explicit
null-count calculation for the string offsets column.

Signed-off-by: MithunR <mythrocks@gmail.com>

This is a followup to NVIDIA#1148.

`row_conversion.cu` was modified in rapidsai/cudf#13372
to explicitly calculate null-counts for output columns.

This commit replicates the changes in cudf/pull/13372, and adds explicit
null-count calculation for the string offsets column.

Signed-off-by: MithunR <mythrocks@gmail.com>

* Followup for null count fixup in row_conversion.cu.

This is a followup to #1148.

`row_conversion.cu` was modified in rapidsai/cudf#13372
to explicitly calculate null-counts for output columns.

This commit replicates the changes in cudf/pull/13372, and adds explicit
null-count calculation for the string offsets column.

Signed-off-by: MithunR <mythrocks@gmail.com>

This reverts commit 56150d9.

Adds the `null_count()` member function back to the `cudf::column_device_view` class.
This member function was removed in #4799 because the null count was unreliable since it could be set to `UNKNOWN_NULL_COUNT`. The `UNKNOWN_NULL_COUNT`  was removed in #13372 so this should no longer be an issue.

Although the value is now more reliable it is just a copy at the time the `cudf::column_device_view` was created so it is not without caveats though these are easily explained and hopefully understood. The value is much less reliable on `cudf::mutable_column_device_view` where the null-mask is more likely modified rendering the count invalid. Therefore, the `null_count()` is added only to the immutable `cudf::column_device_view`.

The growing number of changes to move template parameters like `has_nulls` to runtime parameters reduces unneeded kernel compilations. Here is one example where having this member function would've helped to reduce the function parameters and more clearly identify the `has_nulls` parameter directly correlates to the `input` parameter state:
https://github.com/rapidsai/cudf/blob/609b08e2c8075cad12347cede69d195fa5b186f1/cpp/src/rolling/detail/rolling_operators.cuh#L399-L416

Note the `has_nulls` parameter would not be needed since the `input` could contain this information and more clearly show what the parameter is based on.

Authors:
  - David Wendt (https://github.com/davidwendt)

Approvers:
  - Tianyu Liu (https://github.com/kingcrimsontianyu)
  - Nghia Truong (https://github.com/ttnghia)
  - Muhammad Haseeb (https://github.com/mhaseeb123)

URL: #21430