Implement __array__() method for DNDarrays

[ClaudiaComito]

Related
#1031

Feature functionality
Provide a view of the process-local array (a slice of the global array) as a numpy ndarray.

This is very different from the dndarray.numpy() method that provides a copy of the global array (gathered on each MPI process) as a numpy ndarray.

Potential confusion will be addressed in the documentation.

Additional context
Required for interoperability with the numpy ecosystem including the xarray library. Missing __array__ method results in numpy not recognizing the DNDarray as an array object, e.g.:

>>> import heat as ht
>>> import numpy as np
>>> a = ht.arange(10)
>>> a
DNDarray([0, 1, 2, 3, 4, 5, 6, 7, 8, 9], dtype=ht.int32, device=cpu:0, split=None)
>>> np.asarray(a)
array([DNDarray(0, dtype=ht.int32, device=cpu:0, split=None),
       DNDarray(1, dtype=ht.int32, device=cpu:0, split=None),
       DNDarray(2, dtype=ht.int32, device=cpu:0, split=None),
       DNDarray(3, dtype=ht.int32, device=cpu:0, split=None),
       DNDarray(4, dtype=ht.int32, device=cpu:0, split=None),
       DNDarray(5, dtype=ht.int32, device=cpu:0, split=None),
       DNDarray(6, dtype=ht.int32, device=cpu:0, split=None),
       DNDarray(7, dtype=ht.int32, device=cpu:0, split=None),
       DNDarray(8, dtype=ht.int32, device=cpu:0, split=None),
       DNDarray(9, dtype=ht.int32, device=cpu:0, split=None)],
      dtype=object)

whereas interoperability requires:

>>> np.asarray(a)
array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9], dtype=int32)