dask.pad does not work with cupy arrays

The following throws an exception with `cupy`, but works fine with `numpy`.

```
import numpy as np
import cupy as cp

import dask.array as da

x = cp.random.randn(128, 128, 128)

da.pad(x, tuple((0, x.shape[dim]) for dim in range(x.ndim)), mode = 'constant')
```

Note that `cupy` supports the pad function.