new domains new classes continual learning and deprecated functions

[chiaralansia]

Thank you for the hard work behind it and for making it open source!

So far, I’ve been working with (and actively modifying to fit my needs) an older version of the library.
I'm currently updating my code to the latest version to benefit from all the fixes and improvements.

I’ve noticed that, for benchmarks with standard datasets (MNIST, CIFAR, etc.), you’re still using functions like nc_benchmark, which are marked as deprecated in the updated folders.

In my case, I had created a function called nc_nd_benchmark to support different domains that share common classes — meaning that classes are also repeated across the experience stream.

My question is: since nc_benchmark is considered deprecated, should I update my code to use the new CLExperience/class_incremental_benchmark approach?
Is there a function or utility in the library that already handles different domains with overlapping classes? (I might have missed it.)
Or would you recommend sticking to the deprecated structure, given that standard dataset management in the current library still seems to rely on it?

Thanks!!