This pull request introduces a new data augmentation module for time series, specifically frequency domain perturbation, and enhances the WienerFilterSimulator with improved property management and testability. It also adds comprehensive unit tests for both the new augmentation functionality and the Wiener filter simulator. The package version is incremented to 0.0.8.

New Features and Enhancements:

Data Augmentation:

• Added a new augmentation module with a frequency_perturbation function for applying frequency domain perturbations to time series data, along with a helper function sample_random_perturbation. This allows users to augment data for improved model robustness. (s2generator/augmentation/__init__.py, s2generator/augmentation/frequency_perturbation.py, s2generator/__init__.py) [1] [2] [3]
• Introduced unit tests for the augmentation module to validate the correctness of perturbation sampling and application. (tests/test_augmentation.py)

Wiener Filter Simulator Improvements:

• Refactored the WienerFilterSimulator class to use a private attribute _sigma_sq for noise variance, and added public getter property sigma_sq for safer access. (s2generator/simulator/wiener_filter.py) [1] [2] [3]
• Added set_coeffs and set_sigma_sq methods to allow manual setting of filter coefficients and noise variance, facilitating easier testing and customization. (s2generator/simulator/wiener_filter.py)
• Updated the simulator's __all__ for proper exports. (s2generator/simulator/__init__.py)
• Added comprehensive unit tests for the WienerFilterSimulator, covering instantiation, fitting, transformation, property access, and manual parameter setting. (tests/test_wiener_filter_simulator.py)

Other Changes:

Versioning:

• Bumped package version from 0.0.7 to 0.0.8 in both s2generator/__init__.py and setup.py to reflect new features and improvements. [1] [2]