GPU compatible (L-)C2ST

[JuliaLinhart]

Is your feature request related to a problem? Please describe.
For high-dimensional datasets, running C2ST (and especially L-C2ST) can be fairly slow. Unfortunately, these metrics / diagnostics rely on the use of sklearn classifiers that are not GPU compatible.

Describe the solution you'd like
It would be nice to allow the use of classifiers from other libraries, such as skorch, that can be trained with a GPU.

Describe alternatives you've considered
NA

Additional context
NA