Fix OverflowError: Python int too large to convert to C long

[kounelisagis]

The problem is related to https://numpy.org/neps/nep-0050-scalar-promotion.html#nep50.
Especially the fact that in numpy2 this behavior is not valid anymore:

>>> np.uint64(5) + -1
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
OverflowError: Python integer -1 out of bounds for uint64

And we have to do this:

>>> np.uint64(5) + np.int64(-1)
np.float64(4.0)

I had made a fix when support for numpy2 was added, converting start,stop,step to np.int64, but this caused this overflow.

[kounelisagis]

@jparismorgan could you confirm this fixes your original issue?

[jparismorgan]

Thanks for fixing this!