Fix segmented_reduce on empty column with non-empty offsets

[davidwendt]

Fixes cudf::segmented_reduce where the input values column is empty but the offsets are not. In this case, the offsets vector {0,0} specifies an empty segment which should result in a single null row. The logic has been fixed and new gtest cases have been added.

Closes #10556

[codecov]

Codecov Report

Merging #10876 (6783b90) into branch-22.06 (6352b4e) will increase coverage by 0.03%.
The diff coverage is 100.00%.

❗ Current head 6783b90 differs from pull request most recent head 889e4ee. Consider uploading reports for the commit 889e4ee to get more accurate results

@@               Coverage Diff                @@
##           branch-22.06   #10876      +/-   ##
================================================
+ Coverage         86.29%   86.33%   +0.03%     
================================================
  Files               144      144              
  Lines             22656    22665       +9     
================================================
+ Hits              19552    19567      +15     
+ Misses             3104     3098       -6     

Impacted Files	Coverage Δ
python/cudf/cudf/core/_base_index.py	85.71% <ø> (ø)
python/cudf/cudf/core/frame.py	93.41% <ø> (ø)
python/cudf/cudf/core/groupby/groupby.py	91.79% <ø> (+0.22%)	⬆️
python/cudf/cudf/core/index.py	92.06% <ø> (ø)
python/cudf/cudf/core/indexed_frame.py	91.70% <ø> (ø)
python/cudf/cudf/core/series.py	95.17% <ø> (ø)
python/cudf/cudf/core/column/string.py	88.78% <100.00%> (+0.12%)	⬆️
python/dask_cudf/dask_cudf/groupby.py	97.42% <100.00%> (+<0.01%)	⬆️
python/dask_cudf/dask_cudf/tests/test_groupby.py	100.00% <100.00%> (ø)
python/cudf/cudf/core/dataframe.py	93.78% <0.00%> (+0.04%)	⬆️
... and 4 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 7f9d51b...889e4ee. Read the comment docs.

[bdice]

Great! I suppose we tried to be too "clever" with the early exit here and it bit us.

[jlowe]

Tested this with the original Spark query that triggered a failure and it now produces the correct results with this change. Thanks, @davidwendt!

[davidwendt]

@gpucibot merge