BUG: np.nan converts to np.int64 without error.

[Micky774]

Describe the issue:

Currently np.array([np.nan], dtype=np.int64) raises an error, however np.array([np.nan]).astype(np.int64) is undefined behavior that only issues a warning. Would it be worth promoting this to an error to stay consistent with np.array behavior?

Reproduce the code example:

import numpy as np

np.array([np.nan]).astype(np.int64) # Works while issuing 'invalid value' warning

np.array([np.nan], dtype=np.int64) # Raises 'ValueError: cannot convert float NaN to integer'

Error message:

No response

NumPy/Python version information:

1.24.0.dev0+578.g918065167 3.9.12 (main, Jun  1 2022, 11:38:51) 
[GCC 7.5.0]

[seberg]

PR that introduced warnings is gh-21437 a duplicate of this that I did not close is gh-14412