ci: tpcds benchmarking

[raunakab]

Overview

This PR introduces a new script called tpcds.py which runs the build-commit workflow (only if a pre-built wheel does not exist), and then runs the run-cluster workflow on the built wheel.

Note

The build-commit.yaml workflow has been changed to not perform any checking if python wheel exists; that logic has been moved into the tpcds.py script.

Usage

uv run tools/tpcds.py --ref $MY_BRANCH --force

[codspeed-hq]

CodSpeed Performance Report

Merging #3597 will improve performances by 42.54%

_{Comparing ci/tpcds (9fba5d2) with main (6602502)}

Summary

⚡ 1 improvements
✅ 26 untouched benchmarks

Benchmarks breakdown

	Benchmark	main	ci/tpcds	Change
⚡	test_iter_rows_first_row[100 Small Files]	236.5 ms	165.9 ms	+42.54%

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 77.93%. Comparing base (cc5ad00) to head (9fba5d2).
Report is 18 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3597      +/-   ##
==========================================
+ Coverage   77.85%   77.93%   +0.07%     
==========================================
  Files         719      720       +1     
  Lines       88455    88480      +25     
==========================================
+ Hits        68871    68960      +89     
+ Misses      19584    19520      -64     

see 29 files with indirect coverage changes

[jaychia]

Logic for checking if the wheel exists/triggering the workflow should not be in TPC-DS workflow.

It's re-usable for other jobs as well (I'd want to use it for my jobs). Likely belongs in the run-cluster workflow, or a wrapper of that which runs one after the other.

[raunakab]

Closing in favour of #3619.