dpnp.cov() doesn't work on Iris XE

[antonwolfy]

dpnp.cov function doesn't work on GPU device without double precision support:

In [1]: import dpnp as np

In [2]: x = np.array([[0, 2], [1, 1], [2, 0]]).T

In [3]: np.cov(x)
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[3], line 1
----> 1 np.cov(x)

File ~/code/dpnp/dpnp/dpnp_iface_statistics.py:295, in cov(x1, y, rowvar, bias, ddof, fweights, aweights)
    293     else:
    294         if x1_desc.dtype != dpnp.float64:
--> 295             x1_desc = dpnp.get_dpnp_descriptor(dpnp.astype(x1, dpnp.float64), copy_when_nondefault_queue=False)
    297         return dpnp_cov(x1_desc).get_pyobj()
    299 return call_origin(numpy.cov, x1, y, rowvar, bias, ddof, fweights, aweights)

File ~/code/dpnp/dpnp/dpnp_iface.py:160, in astype(x1, dtype, order, casting, subok, copy)
    158 """Copy the array with data type casting."""
    159 if isinstance(x1, dpnp_array):
--> 160     return x1.astype(dtype, order=order, casting=casting, copy=copy)
    162 if isinstance(x1, dpt.usm_ndarray):
    163     return dpt.astype(x1, dtype, order=order, casting=casting, copy=copy)

File ~/code/dpnp/dpnp/dpnp_array.py:576, in dpnp_array.astype(self, dtype, order, casting, subok, copy)
    554 """Copy the array with data type casting.
    555
    556 Args:
   (...)
    572
    573 """
    575 new_array = self.__new__(dpnp_array)
--> 576 new_array._array_obj = dpt.astype(self._array_obj, dtype, order=order, casting=casting, copy=copy)
    577 return new_array

File ~/miniconda3/envs/dpnp_dev/lib/python3.9/site-packages/dpctl/tensor/_copy_utils.py:426, in astype(usm_ary, newdtype, order, casting, copy)
    421 else:
    422     raise ValueError(
    423         "Unrecognized value of the order keyword. "
    424         "Recognized values are 'A', 'C', 'F', or 'K'"
    425     )
--> 426 R = dpt.usm_ndarray(
    427     usm_ary.shape,
    428     dtype=target_dtype,
    429     buffer=usm_ary.usm_type,
    430     order=copy_order,
    431     buffer_ctor_kwargs={"queue": usm_ary.sycl_queue},
    432 )
    433 if order == "K" and (not c_contig and not f_contig):
    434     original_strides = usm_ary.strides

File dpctl/tensor/_usmarray.pyx:326, in dpctl.tensor._usmarray.usm_ndarray.__cinit__()

ValueError: Device Intel(R) Graphics [0x9a49] does not support float64 natively.

[vtavana]

completed by #1385