Fixes #649

```

In [1]: import dpctl.tensor as dpt, itertools

In [2]: a = dpt.usm_ndarray((2,3))

In [3]: for i,j in itertools.product(range(2), range(3)): a[i, j] = i*3 + j

In [4]: dpt.to_numpy(a)[1:].T
Out[4]:
array([[3.],
       [4.],
       [5.]])

In [5]: dpt.to_numpy(a[1:].T)
Out[5]:
array([[3.],
       [4.],
       [5.]])

```

Author: oleksandr-pavlyk

dpctl/tensor/_usmarray.pyx

@@ -1054,7 +1054,8 @@ cdef usm_ndarray _transpose(usm_ndarray ary):
             _make_reversed_int_tuple(ary.nd_, ary.strides_)
             if (ary.strides_) else None),
         buffer=ary.base_,
-        order=('F' if (ary.flags_ & USM_ARRAY_C_CONTIGUOUS) else 'C')
+        order=('F' if (ary.flags_ & USM_ARRAY_C_CONTIGUOUS) else 'C'),
+        offset=ary.get_offset()
     )
     r.flags_ |= (ary.flags_ & USM_ARRAY_WRITEABLE)
     return r