silent truncation of numpy int64 arrays to int32 range

[mattjj]

Description

import jax
import numpy

y = jax.jit(lambda x: x + 0)(numpy.array([2**32]))
print(y)  # Array([0])

Consider fixing!

#15275 was an attempt, but it bitrotted.

What jax/jaxlib version are you using?

No response

Which accelerator(s) are you using?

No response

Additional system info

No response

NVIDIA GPU info

No response

[jakevdp]

I believe this is related to the TODO here: https://github.com/google/jax/blob/1c1dd7c8c7ff2e5790159d9cdbbbb1f029a92d4b/jax/_src/numpy/lax_numpy.py#L2066-L2070