[REVIEW] Create agg() function for dataframes

[skirui-source]

Closes #5247

Adds agg function for DataFrame

[GPUtester]

Please update the changelog in order to start CI tests.

View the gpuCI docs here.

[codecov]

Codecov Report

Merging #6483 (5a8fbc1) into branch-0.17 (5336301) will decrease coverage by 0.01%.
The diff coverage is 80.00%.

@@               Coverage Diff               @@
##           branch-0.17    #6483      +/-   ##
===============================================
- Coverage        81.96%   81.95%   -0.02%     
===============================================
  Files               96       96              
  Lines            16177    16294     +117     
===============================================
+ Hits             13260    13353      +93     
- Misses            2917     2941      +24     

Impacted Files	Coverage Δ
python/cudf/cudf/core/dataframe.py	90.77% <80.00%> (-0.38%)	⬇️
python/cudf/cudf/io/json.py	96.55% <0.00%> (-3.45%)	⬇️
python/cudf/cudf/io/avro.py	78.57% <0.00%> (-3.25%)	⬇️
python/cudf/cudf/io/csv.py	94.00% <0.00%> (-1.75%)	⬇️
python/cudf/cudf/utils/ioutils.py	78.71% <0.00%> (-1.18%)	⬇️
python/cudf/cudf/io/orc.py	88.40% <0.00%> (-0.99%)	⬇️
python/cudf/cudf/utils/dtypes.py	89.10% <0.00%> (+0.38%)	⬆️
python/cudf/cudf/io/parquet.py	92.12% <0.00%> (+0.45%)	⬆️
python/cudf/cudf/utils/utils.py	85.35% <0.00%> (+0.67%)	⬆️

---

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 5336301...5a8fbc1. Read the comment docs.

[isVoid]

As we discussed, pd.agg does type promotions to the aggregated results:

>>> pdf = pd.DataFrame({"a":[1,2,3], "b":[3.0, 4.0, 5.0], "c":[True, True, False]})
>>> pdf.agg("sum")
a     6.0
b    12.0
c     2.0
dtype: float64

>>> pdf = pd.DataFrame({"a":[1,2,3], "b":[3, 4, 5], "c":[True, True, False]})
>>> pdf.agg("sum")
a     6
b    12
c     2
dtype: int64

I would suggest you use np.find_common_type() to find out about the promoted dtypes. And later when #6415 merges you could switch to cudf.utils.dtype.find_common_type

[kkraus14]

FYI I changed the branch to 0.17 so you'll likely need to merge in upstream

[harrism]

I think this needs "feature", "improvement", or "bug" label (and to pass CI).

[kkraus14]

Looks like a few failures need to be addressed: https://gpuci.gpuopenanalytics.com/job/rapidsai/job/gpuci-a100/job/cudf/job/prb/job/cudf-gpu-build/2204/testReport/

[rgsl888prabhu]

@skirui-source test cases are failing