IJCNN1 nonlinear svm model

[Vanadjy]

Adding a new problem to RegularizedOptimization : nonlinear SVM problem IJCNN1.

This is a new problem taken from LIBSVM which is compatible with the current structure of generating nonlinear svm problems. Like all other LIBSVM problem using the same activation function, it only requires for data to be loaded and use svm_model on it to generate an instance of the models.

What do you think about it @dpo, @MohamedLaghdafHABIBOULLAH ?

[dpo]

Please add unit tests. It looks like you're using modules that are not current dependencies.

[dpo]

@Vanadjy Please do not commit Manifest.toml.

[dpo]

@Vanadjy ⬆️

Also, your branch is out of sync. Could you please rebase?

[MohamedLaghdafHABIBOULLAH]

Thanks, @Vanadjy. Could you please rebase and sync this PR so we can proceed with the review?
I have a few general comments:

• Remove Manifest.toml (as we can recreate your environment using instantiate)
• It seems you've included some new packages needed conditionally for the IJCNN1 problem. I suggest handling them similarly tor fh_model.
• Move these packages from deps to extras in Project.toml.

[dpo]

Bump @Vanadjy. ☝️