REF: remove groupby_helper

[jbrockmendel]

no longer necessary now that the funcs there use fused types.

[WillAyd]

Can you run benchmarks to make sure no difference?

[jbrockmendel]

Can you run benchmarks to make sure no difference?

Sure, running now. But this is really benign, moving code that is already #included directly into the file where it already lives post-include.

[WillAyd]

For sure. Those benchmarks also take a while...but for moving a large chunk like this better safe than sorry

[jbrockmendel]

Noise

       before           after         ratio
     [da3d0d92]       [c601d34f]
     <master>         <gbh>     
+      11.1±0.5ms       15.1±0.3ms     1.36  groupby.Nth.time_series_nth('datetime')
+      8.46±0.3ms       9.94±0.3ms     1.18  groupby.AggFunctions.time_different_numpy_functions
-        137±10μs        122±0.5μs     0.89  groupby.GroupByMethods.time_dtype_as_group('int', 'shift', 'direct')
-      2.27±0.2ms      2.01±0.01ms     0.89  groupby.GroupManyLabels.time_sum(1)
-        306±20μs          268±3μs     0.88  groupby.GroupByMethods.time_dtype_as_field('float', 'std', 'transformation')
-      1.10±0.1ms          958±5μs     0.87  groupby.GroupByMethods.time_dtype_as_field('float', 'value_counts', 'direct')
-        293±30μs          230±3μs     0.78  groupby.GroupByMethods.time_dtype_as_group('int', 'min', 'transformation')
-      1.70±0.3ms      1.33±0.02ms     0.78  groupby.SumBools.time_groupby_sum_booleans
-        531±80μs         408±10μs     0.77  groupby.GroupByMethods.time_dtype_as_group('float', 'sum', 'transformation')
-        435±30μs          332±9μs     0.76  groupby.GroupByMethods.time_dtype_as_field('float', 'head', 'direct')

SOME BENCHMARKS HAVE CHANGED SIGNIFICANTLY.

[jreback]

thanks!