Move lists utility function definition out of header

[mythrocks]

Fixes #7265.

cudf::detail::get_num_child_rows() is currently defined in cudf/lists/detail/utilities.cuh. The build pipelines for #7189 are fine, but there seem to be build failures in dependent projects such as spark-rapids:

[2021-01-31T08:12:10.611Z] /.../workspace/spark/cudf18_nightly/cpp/include/cudf/lists/detail/utilities.cuh:31:18: error: 'cudf::size_type cudf::detail::get_num_child_rows(const cudf::column_view&, rmm::cuda_stream_view)' defined but not used [-Werror=unused-function]
[2021-01-31T08:12:10.611Z]  static cudf::size_type get_num_child_rows(cudf::column_view const& list_offsets,
[2021-01-31T08:12:10.611Z]                   ^~~~~~~~~~~~~~~~~~
[2021-01-31T08:12:11.981Z] cc1plus: all warnings being treated as errors
[2021-01-31T08:12:12.238Z] make[2]: *** [CMakeFiles/cudf_hash.dir/build.make:82: CMakeFiles/cudf_hash.dir/src/hash/hashing.cu.o] Error 1
[2021-01-31T08:12:12.238Z] make[1]: *** [CMakeFiles/Makefile2:220: CMakeFiles/cudf_hash.dir/all] Error 2

In any case, it is less than ideal for the function to be completely defined in the header, especially given that the likes of hashing.cu are exposed to it (by way of scatter.cuh).

This commit moves the function definition to a separate translation unit, without changing implementation or interface.

[codecov]

Codecov Report

Merging #7266 (710a2cf) into branch-0.18 (8860baf) will increase coverage by 0.11%.
The diff coverage is n/a.

@@               Coverage Diff               @@
##           branch-0.18    #7266      +/-   ##
===============================================
+ Coverage        82.09%   82.20%   +0.11%     
===============================================
  Files               97      100       +3     
  Lines            16474    16952     +478     
===============================================
+ Hits             13524    13936     +412     
- Misses            2950     3016      +66     

Impacted Files	Coverage Δ
python/cudf/cudf/__init__.py	100.00% <ø> (ø)
python/cudf/cudf/_fuzz_testing/avro.py	0.00% <ø> (ø)
python/cudf/cudf/_fuzz_testing/csv.py	0.00% <ø> (ø)
python/cudf/cudf/_fuzz_testing/fuzzer.py	0.00% <ø> (ø)
python/cudf/cudf/_fuzz_testing/io.py	0.00% <ø> (ø)
python/cudf/cudf/_fuzz_testing/json.py	0.00% <ø> (ø)
python/cudf/cudf/_fuzz_testing/main.py	0.00% <ø> (ø)
python/cudf/cudf/_fuzz_testing/orc.py	0.00% <ø> (ø)
python/cudf/cudf/_fuzz_testing/parquet.py	0.00% <ø> (ø)
python/cudf/cudf/_fuzz_testing/utils.py	0.00% <ø> (ø)
... and 83 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 900c1e1...3fb4965. Read the comment docs.

[davidwendt]

Just an observation. Neither the .cuh nor the .cu file have any device code in them. So I think these could just be .hpp and .cpp files respectively instead.
You may need to include cuda_runtime.h in the .cpp file for the cudaMemcpyAsync declaration.

[mythrocks]

@davidwendt, I've switched these to .hpp/.cpp. I've retested.

[hyperbolic2346]

Replacing new code with existing, identical code. I like your style!

[mythrocks]

Letting @vuule, @karthikeyann, and @rapidsai/ops-codeowners off the hook on this one.

Thanks for the reviews, @davidwendt, @nvdbaranec, @hyperbolic2346!

[mythrocks]

rerun tests

[mythrocks]

rerun tests

[mythrocks]

@kkraus14, could this PR please be merged? I'm afraid the spark-rapids builds are broken without this change.
I'm not sure why it didn't merge automatically after the tests ran.

[kkraus14]

@gpucibot merge

[mythrocks]

@kkraus14, Thank you! Wow, that was quick.