odlcuda broken by tensor branch

[adler-j]

I'll see if I can get it back alive, with that said the cupy approach seems far supperior.

[mehrhardt]

Any news on this front? Is odlcuda still the preferred approach for cuda in odl?

[adler-j]

cupy will be the preferred approach once I manage to get it in (and frankly I'm quite close so I should focus on it).

[mehrhardt]

I am looking forward to this one!