bench: Add benchmarks for ordered ARRAY_AGG and STRING_AGG

[lyne7-sc]

Which issue does this PR close?

• Related to Very high memory usage for unnest + group by #20788.

Rationale for this change

Add benchmarks for ordered aggregate functions, including the ordered ARRAY_AGG workload

What changes are included in this PR?

This PR adds a reusable SQL benchmark suite for aggregate functions.

The initial cases cover:

• ARRAY_AGG with multiple ordering expressions.
• Two ARRAY_AGG expressions with different ordering requirements.
• STRING_AGG with multiple ordering expressions.

Are these changes tested?

Yes.

Are there any user-facing changes?

No. This PR only adds benchmarks.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.19%. Comparing base (1b67f2e) to head (5991efe).
⚠️ Report is 4 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #24391      +/-   ##
==========================================
- Coverage   81.19%   81.19%   -0.01%     
==========================================
  Files        1110     1110              
  Lines      388616   388750     +134     
  Branches   388616   388750     +134     
==========================================
+ Hits       315529   315627      +98     
- Misses      54506    54535      +29     
- Partials    18581    18588       +7     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[alamb]

Thanks @lyne7-sc this is a neat idea to introduce a new benchmark suite. I wonder if there are any existing suites that might be better places to add this rather than a whole new suite. Or if we are going to make a new suite, shall we consolidate other benchmarks here for aggregates 🤔

FYI @Omega359 -- here is a use of the new sql benchmark suite

[Omega359]

Thanks @alamb - it's good to know that it's being used :) I do have two PR's in progress to sync up the benchmarks with the latest changes and to impl the bench.sh linkage. I just need to find the time which is in short supply given the amazing weather right now :)

[alamb]

Thanks @alamb - it's good to know that it's being used :) I do have two PR's in progress to sync up the benchmarks with the latest changes and to impl the bench.sh linkage. I just need to find the time which is in short supply given the amazing weather right now :)

Nice -- will look forward to them (and I totally understand re weather -- it is pretty nice here too!)

[lyne7-sc]

Thanks @alamb and @Omega359. The new SQL benchmark framework is very nice to use.

Most aggregate benchmarks I found focus on individual accumulator operations such as update_batch or merge_batch, making it harder to observe an aggregate’s performance and memory behavior across the complete query execution path. I also considered aggregate_query_sql.rs, but benchmarking aggregate functions often requires a broad range of workloads and data distributions, and the new SQL benchmark framework seems more flexible for defining and extending those cases.

If this direction makes sense, I think we could migrate the other end-to-end aggregate cases into this suite. Alternatively, if there is another existing suite or location that would be a better fit, I’m happy to move these cases there.