add null order support to detail::drop_duplicates

[cwharris]

Fixes #6144

Saves us a couple of kernel calls inside cudf::encode by ensuring nulls are always in the expected order.

[codecov]

Codecov Report

Merging #7938 (84b0fc1) into branch-0.20 (599f62d) will increase coverage by 0.42%.
The diff coverage is 92.73%.

❗ Current head 84b0fc1 differs from pull request most recent head b0d2696. Consider uploading reports for the commit b0d2696 to get more accurate results

@@               Coverage Diff               @@
##           branch-0.20    #7938      +/-   ##
===============================================
+ Coverage        82.30%   82.72%   +0.42%     
===============================================
  Files              101      103       +2     
  Lines            17053    17714     +661     
===============================================
+ Hits             14035    14654     +619     
- Misses            3018     3060      +42     

Impacted Files	Coverage Δ
python/cudf/cudf/utils/dtypes.py	83.44% <ø> (-6.45%)	⬇️
python/cudf/cudf/utils/utils.py	83.25% <ø> (-1.81%)	⬇️
python/dask_cudf/dask_cudf/backends.py	89.58% <ø> (-0.05%)	⬇️
python/cudf/cudf/core/groupby/groupby.py	92.41% <78.57%> (-1.04%)	⬇️
python/cudf/cudf/core/column/lists.py	87.41% <80.00%> (+0.19%)	⬆️
python/cudf/cudf/core/column/struct.py	96.29% <86.66%> (-3.71%)	⬇️
python/cudf/cudf/core/index.py	93.04% <88.09%> (+0.01%)	⬆️
python/cudf/cudf/core/column/column.py	87.86% <88.57%> (+0.43%)	⬆️
python/cudf/cudf/core/column/decimal.py	92.92% <91.48%> (-0.92%)	⬇️
python/cudf/cudf/core/column/interval.py	91.11% <92.30%> (+0.48%)	⬆️
... and 67 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 348ad4d...b0d2696. Read the comment docs.

[karthikeyann]

Documentation is not updated yet for the additional parameter.

[davidwendt]

Looks good to me.

[ttnghia]

I wonder if this drop_duplicates API has any usage problem with NaN handling similar to lists::drop_list_duplicates?
Reference: #7662

[cwharris]

@gpucibot merge