Wrap additional attacks from imported libraries

[maurapintor]

Currently, secml-torch provides wrappers for a limited set of attacks from the various libraries it imports. This may restrict the flexibility and benchmarking capabilities for end users who wish to experiment with a wider range of adversarial attack methods.

Proposed Solution

• Identify additional attacks available in the libraries already imported by secml-torch (e.g., Foolbox, Adversarial Library, etc.).
• Implement wrappers for these attacks, following the conventions and structure used in the existing codebase.
• Ensure new wrappers are well documented and tested.
• Update documentation to include usage examples for the newly wrapped attacks.
  • Include documentation on how to wrap attacks (so that users can potentially wrap attacks that are not explicitly wrapped by SecML-Torch Tutorial: wrapping attacks from other libraries #150.