Raise an `IndexError` when array index is not integer

[vtavana]

When index of an array is a floaing point number, NumPy raises an IndexError while dpctl returns an incorrect result.

>>> import numpy, dpctl, dpctl.tensor as dpt
>>> dpctl.__version__
'0.18.0dev0+235.gd79dae1d2f'

>>> numpy.__version__
'1.26.4'

>>> b=dpt.asarray([1+1j, 2+2j])
>>> b[0.].imag
usm_ndarray([])
 
>>> b[1.].imag
usm_ndarray([[1., 2.]])
 
>>> a=numpy.array([1+1j, 2+2j])
>>> a[0.].imag
IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices
 
>>> a[1.].imag
IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices

[oleksandr-pavlyk]

The slicing treats floating points as boolean entities, i.e. 0. is treated as False, while 1., or other non-zero values are treated as True.