feat: More sophisticated evaluation logic

[robinholzi]

Motivation

• Analytics tools showed performance issues when run with multiple big logfiles.
• Naming ambiguities in plotting logic
• No support for averaging over multiple runs

Changes

• Introduce an aggregation tool that transforms multiple pipeline logs into one aggregated (averaged) logfile.
• move analytics plotting app dataframe state into global variable to not reload/rebuild data every time someone selects another pipeline.
• Improve performance of dataframe construction from logfile by not concatenating size 1 dataframes but collecting all logs in tuples and then creating dataframes from them.

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 89.65517% with 6 lines in your changes missing coverage. Please review.

Project coverage is 82.30%. Comparing base (4d78544) to head (dbfbb80).

Files	Patch %	Lines
...yn/supervisor/internal/pipeline_executor/models.py	88.88%	6 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #534 +/- ## ========================================== - Coverage 82.30% 82.30% -0.01% ========================================== Files 217 217 Lines 10118 10146 +28 ========================================== + Hits 8328 8351 +23 - Misses 1790 1795 +5 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[github-actions[bot]]

^{( % to main) ( % to main)}

[robinholzi]

Seems to work smoothly, for now :-) From my side, we can merge this. Thanks so much for the hard work here, this makes analyzing the results so much easier.

Thanks @MaxiBoether, great to hear that!