Adding support for QuoVadis models

[zangobot]

Include wrappers for QuoVadis, by leveraging the fusion models provided inside the original repository by @dtrizna.
It should be implemented as a CQuoVadisClassifier and its wrapper for blackbox attack CQuoVadisWrapperPhi.

The only problem is data preprocessing, as the original code leverage file paths (and later it extracts binaries from them).

[dtrizna]

Note on: The only problem is data preprocessing, as the original code leverage file paths (and later it extracts binaries from them).

In all Speakeasy API documentation, it was defined to use filepath of PE, but I see that .load_module() can take both path= (path) and data= (bytes) as parameters:
https://github.com/mandiant/speakeasy/blob/master/speakeasy/speakeasy.py#L190

I will consider updating QuoVadis API so it allows specifying bytes in preprocessing stage.

[zangobot]

Ok thank you!
I will look into that, as it would be amazing to have QuoVadis inside SecML malware!