FIX-#5545: Align error-handling logic with pandas for groupby.skew

[dchigarev]

Signed-off-by: Dmitry Chigarev dmitry.chigarev@intel.com

What do these changes do?

The PR aligns modin's behavior with pandas' logic on how to behave on invalid results for groupby.skew.

• first commit message and PR title follow format outlined here

  NOTE: If you edit the PR title to match this format, you need to add another commit (even if it's empty) or amend your last commit for the CI job that checks the PR title to pick up the new PR title.

• passes flake8 modin/ asv_bench/benchmarks scripts/doc_checker.py
• passes black --check modin/ asv_bench/benchmarks scripts/doc_checker.py
• signed commit with git commit -s
• Resolves groupby.skew on Linux produces different from pandas results in case of incorrect input #5545
• tests added and passing
• module layout described at docs/development/architecture.rst is up-to-date

groupby.skew() performance hasn't changed significantly:

ASV benchmarks

Steps to run the benchmarks were taken from #5318

       before           after         ratio
     [53d9096d]       [959a3eac]
     <master>       <issue-5545>
         338±30ms         358±20ms     1.06  benchmarks.TimeGroupByMultiColumn.time_groupby_agg_skew([1000000, 32], 'huge_amount_groups', 6)
        1.64±0.2s       1.74±0.06s     1.06  benchmarks.TimeGroupByMultiColumn.time_groupby_agg_skew([1000000, 256], 'huge_amount_groups', 6)
        1.79±0.1s       1.89±0.08s     1.06  benchmarks.TimeGroupByMultiColumn.time_groupby_agg_skew([10000000, 32], 100, 6)
         871±20ms         915±50ms     1.05  benchmarks.TimeGroupByMultiColumn.time_groupby_agg_skew([5000, 5000], 'huge_amount_groups', 6)
          383±6ms          402±8ms     1.05  benchmarks.TimeGroupByDefaultAggregations.time_groupby_skew([5000, 5000], 100)
         196±20ms         205±20ms     1.04  benchmarks.TimeGroupByMultiColumn.time_groupby_agg_skew([1000000, 32], 100, 6)
         790±20ms         818±30ms     1.04  benchmarks.TimeGroupByDefaultAggregations.time_groupby_skew([5000, 5000], 'huge_amount_groups')
         189±10ms         196±20ms     1.03  benchmarks.TimeGroupByDefaultAggregations.time_groupby_skew([1000000, 32], 100)
       1.92±0.07s       1.93±0.07s     1.01  benchmarks.TimeGroupByDefaultAggregations.time_groupby_skew([10000000, 32], 'huge_amount_groups')
       1.34±0.03s       1.34±0.02s     1.00  benchmarks.TimeGroupByMultiColumn.time_groupby_agg_skew([1000000, 256], 100, 6)
         474±20ms         472±20ms     1.00  benchmarks.TimeGroupByMultiColumn.time_groupby_agg_skew([5000, 5000], 100, 6)
       1.70±0.06s       1.67±0.06s     0.98  benchmarks.TimeGroupByDefaultAggregations.time_groupby_skew([10000000, 32], 100)
       1.25±0.05s       1.23±0.03s     0.98  benchmarks.TimeGroupByDefaultAggregations.time_groupby_skew([1000000, 256], 100)
       1.52±0.07s       1.48±0.04s     0.97  benchmarks.TimeGroupByDefaultAggregations.time_groupby_skew([1000000, 256], 'huge_amount_groups')

BENCHMARKS NOT SIGNIFICANTLY CHANGED.

[dchigarev]

this logic was taken from pandas

[anmyachev]

LGTM!