Biopsykit

Changes
- `biopsykit.classification.model_selection.SklearnPipelinePermuter`:
- renamed method `best_pipeline` to `best_hyperparameter_pipeline`. **WARNING: This is a breaking change!**
- added new method to fit classification pipelines and save intermediate results to disk
(`fit_and_save_intermediate`).
- all arguments passed to `fit` and `fit_and_save_intermediate` (except for `X` and `y`) are from now on keyword-only
arguments to avoid confusion. **WARNING: This is a breaking change!**
- `merge_permuter_instances` is now a class method that accepts a list of `SklearnPipelinePermuter` instances or a
list of file paths as argument instead of a method that merges one `SklearnPipelinePermuter` instance with the
current instance. **WARNING: This is a breaking change!**
- added new method `update_permuter` to update an `SklearnPipelinePermuter` instance with new model and parameter
dictionaries that can then be used to fit new pipeline permutations.

Changes
- removed `biopsykit.carwatch_logs` module for processing CARWatch log data (moved to own [`carwatch`](https://github.com/mad-lab-fau/carwatch) package)
Bugfixes
- `biopsykit.classification.model_selection.SklearnPipelinePermuter`: some bugfixes and improvements when computing
additional classification metrics
- additional bugfixes in other functions
Misc
- code changes after switching to ruff as new linter

Added
- `biopsykit.classification.analysis`: Added functions to analyze classification results
- `biopsykit.classification.model_selection.SklearnPipelinePermuter`: Added methods to export classification metrics
as latex tables, to merge to different `SklearnPipelinePermuter` instances, and to compute additional classification
metrics.
- various new util functions
Bugfixes and Improvements
- `biopsykit.classification.model_selection.SklearnPipelinePermuter`: fixing random state now works as intended and
produces reproducible results for real!
- `biopsykit.stats.StatsPipeline`: Improved behavior
- `biopsykit.plotting`: Improved layout
- minor bugfixes

Bugfixes
- `bp.io.biopac`:wrong index name for dataframe export when setting parameter `"index = None"`

Added
- `biopsykit.io`:
- Added functions to directly load aTimeLogger files (`biopsykit.io.load_atimelogger_file`) and to
convert time log files into a dictionary with start/end times (`biopsykit.io.convert_time_log_dict`)
- Added support for loading data acquired from the [Biopac](https://www.biopac.com/) system (`biopsykit.io.biopac`)
- `biopsykit.plotting`: Added function to conveniently plot feature pairs (`biopsykit.plotting.feature_pairplot`),
which is a wrapper function for `seaborn.pairplot`
Improvements
- Adapted code to don't show pandas future warnings anymore
- `biopsykit.plotting.hr_ensemble_plot`: Added "is_relative" argument to style y-axis accordingly
- `biopsykit.stats.StatsPipeline`: Improved latex export for different effect sizes
- `biopsykit.io.nilspod`: Improved handling of loading NilsPods files
- `biopsykit.utils.data_processing`: Added support for data dicts with tuples instead of strings as keys
Bugfixes
- some small bugfixes

Improvements
- `biopsykit.signals.ecg.EcgProcessor`: improved handling of errors occurring during ECG processing
- `biopsykit.plotting`: Improved boxplot behavior due to the use of the new light palettes of
[`fau-colors`](https://github.com/mad-lab-fau/fau_colors).
See the [`StatsPipeline & Plotting Example`](https://biopsykit.readthedocs.io/en/latest/examples/_notebooks/StatsPipeline_Plotting_Example.html)
for further information.
Bugfixes:
- `biopsykit.signals.imu.ActivityCounts`: added timezone support ([PR 31](https://github.com/mad-lab-fau/BioPsyKit/pull/31))
- `biopsykit.signals.imu.RestPeriods`: fixed bug in computing resting periods
(old implementation could lead to classifying a long period of activity as the main rest period.)
([PR 32](https://github.com/mad-lab-fau/BioPsyKit/pull/32))
- `biopsykit.signals.imu.ActivityCounts`: fixed bug(s) in computing activity counts
([PR 33](https://github.com/mad-lab-fau/BioPsyKit/pull/33))
- `biopsykit.stats.StatsPipeline`: fixed bugs in LaTeX export of result tables