Fix ActivationDefense and SpectralSignatures expected flattened bug

[f4str]

Description

Modify the ActivateDefense and SpectralSignatures poisoning defenses to flatten the outputs when calling get_activations() on the estimator.

Additionally, the hack in HuggingFace classifier to use the inputs post-flattening was removed since the defenses will now do the flattening and work with any arbitrary layer.

Fixes #2313

Type of change

Please check all relevant options.

• Improvement (non-breaking)
• Bug fix (non-breaking)
• New feature (non-breaking)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

Testing

Please describe the tests that you ran to verify your changes. Consider listing any relevant details of your test configuration.

• Tests for Activation Defense are unchanged
• Tests for Spectral Signatures are unchanged

Test Configuration:

• OS
• Python version
• ART version or commit number
• TensorFlow / Keras / PyTorch / MXNet version

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• My changes have been tested using both CPU and GPU devices

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (a281f62) 76.18% compared to head (fc7cbcb) 84.09%.

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@              Coverage Diff               @@
##           dev_1.17.0    #2327      +/-   ##
==============================================
+ Coverage       76.18%   84.09%   +7.90%     
==============================================
  Files             327      327              
  Lines           30850    30852       +2     
  Branches         5716     5716              
==============================================
+ Hits            23503    25944    +2441     
+ Misses           5914     3448    -2466     
- Partials         1433     1460      +27     

Files	Coverage Δ
art/defences/detector/poison/activation_defence.py	83.28% <100.00%> (+0.04%)	⬆️
...nces/detector/poison/spectral_signature_defense.py	84.72% <100.00%> (+0.21%)	⬆️

... and 63 files with indirect coverage changes

[beat-buesser]

Hi @f4str Thank you very much for your pull request! It looks good to me.

[f4str]

Due to issues with GitHub actions tests failing, removing the hack in the HuggingFace classifier to use the inputs post-flattening will not be included in this PR. Once the failed unit test is further investigated, the change will be revisited.