dpt.sum with "keepsdims" returns incorrect shape in case one or more axis is equal to 0

[antonwolfy]

The problem comes from gh-1487:

import dpctl.tensor as dpt
import numpy

a = dpt.ones((0, 10))
dpt.sum(a, keepdims=True).shape 
# Out: ()

b = numpy.ones((0, 10))
numpy.sum(a, keepdims=True).shape 
# Out: (1,)