Numpy datasets

[t-kalinowski]

This PR adds a convert argument to all the builtin datasets like dataset_mnist().

If called like dataset_mnist(convert = FALSE), arrays are returned as Numpy arrays. e.g.:

str(dataset_mnist())

List of 2
 $ train:List of 2
  ..$ x: int [1:60000, 1:28, 1:28] 0 0 0 0 0 0 0 0 0 0 ...
  ..$ y: int [1:60000(1d)] 5 0 4 1 9 2 1 3 1 4 ...
 $ test :List of 2
  ..$ x: int [1:10000, 1:28, 1:28] 0 0 0 0 0 0 0 0 0 0 ...
  ..$ y: int [1:10000(1d)] 7 2 1 0 4 1 4 9 5 9 ...

str(dataset_mnist(convert = FALSE))

List of 2
 $ train:List of 2
  ..$ x: <numpy.ndarray shape(60000,28,28), dtype=uint8>
  ..$ y: <numpy.ndarray shape(60000), dtype=uint8>
 $ test :List of 2
  ..$ x: <numpy.ndarray shape(10000,28,28), dtype=uint8>
  ..$ y: <numpy.ndarray shape(10000), dtype=uint8>