test(python): Update benchmark tests

[stinodego]

Changes

• Rewrite the R script for data generation as a Python script
• Use smaller data for benchmark tests. CodSpeed tests should be ran on tiny data, so we are now using 1000 10_000 rows.
• Since the data is small, we don't have to write it to disk for re-use, we can generate it on the fly. Reorganized things a bit to take advantage of this.
• Some tests were marked as benchmark tests but didn't need to be. I marked those as 'release' and moved them to the unit test module.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 80.38%. Comparing base (b4e43d7) to head (f8f988e).
Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #15825      +/-   ##
==========================================
- Coverage   80.38%   80.38%   -0.01%     
==========================================
  Files        1263     1263              
  Lines      165346   165368      +22     
==========================================
+ Hits       132918   132926       +8     
- Misses      32428    32442      +14     

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

[codspeed-hq]

CodSpeed Performance Report

Merging #15825 will improve performances by ×560

_{Comparing bench-python (f8f988e) with main (1c6b7b7)}

Summary

⚡ 2 improvements

🆕 11 new benchmarks
⁉️ 20 dropped benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	main	bench-python	Change
⚡	test_filter1	1,191.4 ms	2.1 ms	×560
⚡	test_filter2	1,988.2 ms	2.8 ms	×720
🆕	test_groupby_h2oai_q1	N/A	8.6 ms	N/A
🆕	test_groupby_h2oai_q10	N/A	32.1 ms	N/A
🆕	test_groupby_h2oai_q2	N/A	19.5 ms	N/A
🆕	test_groupby_h2oai_q3	N/A	8.8 ms	N/A
🆕	test_groupby_h2oai_q4	N/A	7.9 ms	N/A
🆕	test_groupby_h2oai_q5	N/A	7.7 ms	N/A
🆕	test_groupby_h2oai_q6	N/A	50.2 ms	N/A
🆕	test_groupby_h2oai_q7	N/A	8.4 ms	N/A
🆕	test_groupby_h2oai_q8	N/A	10.6 ms	N/A
🆕	test_groupby_h2oai_q9	N/A	107.6 ms	N/A
⁉️	test_h2oai_groupby_q1	5.3 s	N/A	N/A
⁉️	test_h2oai_groupby_q10	40.4 s	N/A	N/A
⁉️	test_h2oai_groupby_q2	12.9 s	N/A	N/A
⁉️	test_h2oai_groupby_q3	9.1 s	N/A	N/A
⁉️	test_h2oai_groupby_q4	6.8 s	N/A	N/A
⁉️	test_h2oai_groupby_q5	7.2 s	N/A	N/A
⁉️	test_h2oai_groupby_q6	14.1 s	N/A	N/A
⁉️	test_h2oai_groupby_q7	8.5 s	N/A	N/A
...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

[ritchie46]

Use smaller data for benchmark tests. CodSpeed tests should be ran on tiny data, so we are now using 1000 rows.

Is 10_000 or 50_000 still acceptable? I feel 1000 is too nano for Polars.

[stinodego]

Is 10_000 or 50_000 still acceptable? I feel 1000 is too nano for Polars.

I think that would be fine. I updated to 10_000 rows, let's see how long each benchmark test takes now.