Workaround for dpnp.linalg.qr()
to run on CUDA
#2265
Merged
dpnp.linalg.qr()
to run on CUDA
#2265