Update scikit-learn to v1.4

[desilinguist]

• Update sklearn version in requirements.txt.
• Properly clean up temporary files in test_output.
• Update test data due to sklearn change.
• Add filter for unnecessary matplotlib messages.
• Fix warnings and filter matplotlib messages.
  • Remove deprecated scale parameter from seaborn calls.
  • Add explicit palette call when plotting time curve since that's now required.
  • Use matplotlib.category msg filter to filter out unnecessary msgs.
• Update pre-commit checks.
• Handle sklearn deprecations.
  • needs_proba and needs_threshold arguments for make_scorer() are deprecated in favor of response_method.
  • LinearSVC and LinearSVR now need to have dual explicitly set to "auto".
  • AdaBoostClassifier now needs to have algorithm` set to "SAMME".
  • Update custom metrics documentation.
  • Update tests and test data.
• Update Gitlab CI python version to 3.10 since sklearn v1.4 drops support for both Python 3.8 and 3.9.

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Comparison is base (bc55306) 95.44% compared to head (442c10c) 95.43%.

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #766 +/- ## ========================================== - Coverage 95.44% 95.43% -0.02% ========================================== Files 30 30 Lines 3688 3701 +13 ========================================== + Hits 3520 3532 +12 - Misses 168 169 +1 ```

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