speedup with to_lists()

[dougbrn]

This PR uses to_lists() in place of to_flat() and apply() for the by=None count_nested case and in reduce to offer a 5-10x speedup.

It does introduce one limitation to reduce in that the nested structures cannot have overlapping column names, this will be less of an issue once #33 is addressed. I propose we wait to resolve this overlap issue until after #33 and just make a separate issue for now.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 95.68%. Comparing base (025ad96) to head (fe20c57).

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #34      +/-   ##
==========================================
+ Coverage   95.63%   95.68%   +0.05%     
==========================================
  Files          14       14              
  Lines         618      626       +8     
==========================================
+ Hits          591      599       +8     
  Misses         27       27              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[github-actions]

Before [025ad96]	After [9844d26]	Ratio	Benchmark (Parameter)
2.36±0.8s	1.73±0.9s	~0.73	benchmarks.time_computation
3.54k	640	0.18	benchmarks.mem_list

Click here to view all benchmarks.

[wilsonbb]

An exciting speedup! Just added a comment suggesting we use a different form of join.