[feat] Support multiple aggregations and statistics at once when querying performance data

[vkarak]

This PR offers the following features:

1. Multiple statistics can be presented at once using the special stats aggregator. These includes mean, standard deviation, and 0.01, 0.05, 0.50, 0.95, 0.99 quantiles.
2. The cmp_spec syntax is updated to support multiple statistics at once, e.g., mean,min,max, and also aggregations on values other than the performance, e.g., min(num_tasks),min(pref) etc.
3. Two new command-line options are added to control the name of the left and right terms in the comparisons. These are the --term-lhs and --term-rhs and are only meaningful with the --performance-compare and --performance-report options.
4. The default name suffixes for left/right terms in comparisons are now (lhs) and (rhs) respectively.
5. The selectors of left and right columns in the cmp spec are now _L and _R instead of _A and _B.

Closes #3436.

Implementation details

The analytics layers is now using Polars through its Python bindings.

Managing the column names was the most challenging part of the implementation, as we wanted to keep a straightforward for selecting test attributes from the command line, specifying the left/right variants quickly and the same time produce tables with meaningful headers (e.g., include the aggregation in the column name, especially now, that multiple aggregations are supported at once). This is handled the the internal _QueryMatch class which replaces the very thin former _Match named tuple.

Todos

• Update documentation
• Update tutorial examples
• --table-format=pretty still uses | as a separator for join_uniq.

Notes

If the CI does not pass for Python < 3.9, we should simply merge this PR after #3583 is addressed.

This PR depends on #3596.

[codecov]

Codecov Report

❌ Patch coverage is 95.12195% with 12 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.39%. Comparing base (bd8278f) to head (4bc7d65).
⚠️ Report is 10 commits behind head on develop.

Files with missing lines	Patch %	Lines
reframe/frontend/reporting/utility.py	94.21%	10 Missing ⚠️
reframe/frontend/reporting/__init__.py	96.66%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #3585      +/-   ##
===========================================
+ Coverage    91.37%   91.39%   +0.02%     
===========================================
  Files           62       62              
  Lines        13492    13520      +28     
===========================================
+ Hits         12328    12357      +29     
+ Misses        1164     1163       -1     

☔ View full report in Codecov by Sentry.
📢 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.

[ekouts]

Haven't finished my testing, but stats seems to be broken for --performance-report:

./bin/reframe -c my_checks/perf_vars.py -r --performance-report "now-1d:now/stats:/"
ReFrame Setup]
  version:           4.10.0-dev.0+cc60ca88
  ....
[==========] Finished on Mon Jan 26 11:38:45 2026+0100
WARNING: failed to generate performance report: "pval (std)" not found
Current session stored with UUID: e0a328f6-0ff3-4eeb-85fb-eb0047f17040
Log file(s) saved in '/var/folders/19/lkdv1y8949g05ynf917f07mm0000gn/T/rfm-a76sgaqq.log'

When the database is disabled, reframe doesn't complain about this, just returns the default results.

[vkarak]

Haven't finished my testing, but stats seems to be broken for --performance-report:

@ekouts I fixed this in d51db4e. Note also that I rebased the PR.

[vkarak]

@ekouts I've added also support for any quantile here: 575314a.

[ekouts]

lgtm!