Revert "Check `split_out` to decide on sorted groupby (#802)"

[charlesbluca]

With https://github.com/rapidsai/cudf/pull/11755 merged in and reflected in our gpuCI images, we should no longer need #802 in for gpuCI to pass.

[codecov-commenter]

Codecov Report

Merging #807 (589ff62) into main (6810f1a) will increase coverage by 0.11%. The diff coverage is n/a.

@@            Coverage Diff             @@
##             main     #807      +/-   ##
==========================================
+ Coverage   77.39%   77.50%   +0.11%     
==========================================
  Files          75       75              
  Lines        4220     4219       -1     
  Branches      767      767              
==========================================
+ Hits         3266     3270       +4     
+ Misses        784      775       -9     
- Partials      170      174       +4     

Impacted Files	Coverage Δ
dask_sql/physical/rel/logical/aggregate.py	89.44% <ø> (-0.06%)	:arrow_down:
dask_sql/_version.py	35.31% <0.00%> (+1.41%)	:arrow_up:

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more

[galipremsagar]

rerun tests

[charlesbluca]

Looks like the reason for failures here is that we aren't picking up the latest 22.10 nightlies with the fix:

cudf                      22.10.00a220920 cuda_11_py39_g0528b38f2b_241    rapidsai-nightly
libcudf                   22.10.00a220920 cuda11_g0528b38f2b_241    rapidsai-nightly

Looking into this now

EDIT:

Looks like 22.10 nightlies are no longer being published - assuming that once 22.12 nightlies become available and we bump to use those we should be unblocked here