ToDevice should allocate buf with size equal to storage length, not shape numel

[coreylowman]

let b = a.broadcast::<Rank2<5, 3>, _>();
assert_eq!(b.array(), [[1.0, 2.0, 3.0]; 5]);

let c = b.broadcast::<Rank4<7, 5, 3, 2>, _>().to_device(&Cpu::default());
assert_eq!(c.array(), [[[[1.0; 2], [2.0; 2], [3.0; 2]]; 5]; 7]);

let d = c.mean::<Rank2<5, 3>, _>();
assert_eq!(d.array(), [[1.0, 2.0, 3.0]; 5]);

let e = dev.tensor([1.0]);
let f = e.broadcast::<Rank2<1, 1>, Axis<1>>();

still results in the very weird assertion error inside cudarc-0.7.0/src/driver/safe.rs:531:9.

Originally posted by @TimerErTim in #451 (comment)

[coreylowman]

Caused by

        let mut buf = std::vec![E::default(); self.shape().num_elements()];
        self.copy_into(&mut buf);
        device.tensor((buf, *self.shape()))

let mut buf = std::vec![E::default(); self.shape().num_elements()]; should use the number of elements of the underlying storage instead.

[coreylowman]

Other issues:

1. Cuda as_vec() returns vec of physical storage, Cpu as_vec() returns vec of actual storage
2. CopySlice for both Cpu/Cuda copy physical storage