Fix `clip` for Numpy 2.0

[ndgrigorian]

clip currently fails for out-of-bounds Python integers due to Numpy 2.0 changes:

In [21]: dpt.clip(dpt.asarray([0, 255], dtype=dpt.uint8), 0, 4550)
---------------------------------------------------------------------------
OverflowError                             Traceback (most recent call last)
Cell In[21], line 1
----> 1 dpt.clip(dpt.asarray([0, 255], dtype=dpt.uint8), 0, 4550)

File ~/repos/dpctl/dpctl/tensor/_clip.py:634, in clip(x, min, max, out, order)
    632     a_max = max
    633 else:
--> 634     a_max = dpt.asarray(max, dtype=max_dtype, sycl_queue=exec_q)
    636 if order == "A":
    637     order = (
    638         "F"
    639         if all(
   (...)
    647         else "C"
    648     )

File ~/repos/dpctl/dpctl/tensor/_ctors.py:649, in asarray(obj, dtype, device, copy, usm_type, sycl_queue, order)
    644     raise ValueError(
    645         f"Converting {type(obj)} to usm_ndarray requires a copy"
    646     )
    647 # obj is a scalar, create 0d array
    648 return _asarray_from_numpy_ndarray(
--> 649     np.asarray(obj, dtype=dtype),
    650     dtype=dtype,
    651     usm_type=usm_type,
    652     sycl_queue=sycl_queue,
    653     order="C",
    654 )

OverflowError: Python integer 4550 out of bounds for uint8

Numpy has been doing the same and has changed this in numpy/numpy#26892

dpctl should implement a similar work-around