Enhance activations testing (#992)

[learner0018]

Implements testing framework for activation functions using reference data generated by PyTorch.

Changes:

• Added generate_activation_data.py to generate .npy reference files for sigmoid, relu, tanh, elu, selu, softplus, and softsign.
• Draft implementation of unit tests in test_specialfunctions_activations.fypp using test-drive.

Current Status:
As discussed in #992, I am encountering a build error linking stdlib_io to the test executable.
Opening this as a Draft to get help with the CMake configuration.

Reference data files generated locally (ignored by git).

[learner0018]

Hi @jalvesz,
I have pushed the updates to integrate the testing framework directly into test_specialfunctions_activations.fypp using test-drive as discussed in the issue.
Key changes implemented:
1 CMakeFix: I added stdlib_io to the dependencies in test/specialfunctions/CMakeLists.txt to resolve the module visibility issue.
2 Test Logic: Implemented the initial test subroutines in test_specialfunctions_activations.fypp.
3 Python Script: Created generate_activation_data.py to handle the reference data generation.

Looking forward to your feedback

[jalvesz]

before reveiwing, could you please first remove unrelated changes: the app folder seems to be something coming from local testing, please revert this. Also, the .f90 file in the src folder should not be tracked here, it is a file created by the fypp preprocessor.

[learner0018]

understood. i will remove the app folder and the generated .f90 file immediately. thanks for the guidance.