Add --extra-metadata support to compare_to command

[amritamishra01]

Summary

This pull request adds support for a new --extra-metadata argument to the compare_to command. It allows users to include selected metadata keys as additional columns in the comparison table output.

Motivation

The current compare_to output does not show metadata values such as CPU model, OS version, Python build info, etc. This feature was requested in Issue #162 (“Display additional information through table”) to make analysis more informative.

What This PR Does

Adds a new CLI flag: --extra-metadata "key1,key2,..."

Passes metadata keys into _compare.py

Updates table generation to show metadata for both reference and changed benchmarks

Ensures geometric mean rows remain aligned

Testing

Manually tested using sample benchmark files

Verified table output for:

No metadata requested

Single metadata key

Multiple metadata keys

Mixed significant / non-significant benchmarks

Notes

Feedback is welcome! I’m happy to update the PR if changes are needed.

[amritamishra01]

Hi! @vstinner I’ve pushed the updated changes to the add-extra-metadata branch. Here’s what I did:

Integrated the --extra-metadata feature cleanly into the existing compare_to flow.

Fixed implementation issues that caused errors when formatting metadata output.

Resolved merge conflicts during branch sync and ensured the final code reflects only the correct logic.

Added a dedicated test file: pyperf/tests/test_extra_metadata.py

Verified full test suite — all tests pass successfully (174 passed).

Ensured that:

Baseline behavior of compare_to is unchanged

Metadata output only appears when explicitly requested

No regressions in table, CLI, verbose, or non-significant cases.

Updated the PR branch cleanly using cherry-pick without opening a new PR.

Happy to make any additional adjustments if required!

[amritamishra01]

Hi @vstinner , thanks for the review!

I have addressed all your comments:

Moved the test to pyperf/tests/test_perf_cli.py and used the existing create_bench helper.

Fixed the merge conflict in main.py and cleaned up the argument definition.

Fixed the code style in _compare.py (indentation, comments, and line wrapping) as requested.

Tests are passing locally. Please let me know if there is anything else!