[FEAT] [New Query Planner] Groupby support, aggregation fixes, support for remaining aggregation ops

[clarkzinzow]

This PR adds support for df.groupby(), fixes misc. things with aggregations, and adds support for the remaining (non-sum) aggregation ops.

This PR builds off of #1257, where @xcharleslin implemented the core meat of this PR (this PR just wires things together and fixes a few minor things). From that PR's description:

"Ported over the logic in our existing AggregationPlanBuilder. Groupby-aggregates should now be fully supported (including multi-partition).

Additionally, this PR improves on Daft's existing aggregation logic by using semantic IDs in intermediate results, so that redundant intermediates are not computed.

E.g. before, getting the Sum and Mean of a column would compute and carry around two copies of the intermediate sum, one for the Sum and one for the Mean. Now, all stages address their required intermediates by semantic ID, eliminating these duplicates."

[codecov]

Codecov Report

Merging #1272 (3fcf3f6) into main (c43c76c) will increase coverage by 0.09%.
The diff coverage is 82.35%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1272      +/-   ##
==========================================
+ Coverage   87.66%   87.75%   +0.09%     
==========================================
  Files          61       61              
  Lines        6009     6021      +12     
==========================================
+ Hits         5268     5284      +16     
+ Misses        741      737       -4     

Files Changed	Coverage Δ
daft/logical/rust_logical_plan.py	89.03% <78.57%> (+3.71%)	⬆️
daft/execution/rust_physical_plan_shim.py	98.27% <100.00%> (ø)