TYP: invalid default value documented on array's ndmax parameter #30159

Describe the issue:

Hello from downstream, @jorenham !

first of: we also have a couple inspect-based tests in astropy intended to catch discrepancies between numpy signatures and our NEP 18 dispatch functions. Functions which didn't have signatures available at runtime were simply ignored, but your recent work has allowed finding many more issues with our NEP 18 implementations, so thanks a lot for that !

I'm seeing a small issue with gh-30138 though; None doesn't appear to be a valid value for array's ndmax argument at runtime (an integer is expected). Instead, and based on my understanding of gh-29569, it seems the default value should actually be 0.
It can be worked around in astropy alright, but it feels slightly off that I need to both re-expose the default value as None (granted, this part is imposed by our tests) and include logic in the implementation to handle this special case. It's also a bit awkward that the parameter is typed as int but uses None as a default.

Reproduce the code example:

# a script will not cut it

Error message:

Python and NumPy Versions:

Python 3.11.18 (irrelevant)
numpy 2.4.0dev (d8e1c8c)

Type-checker version and settings:

not actually detected at typecheck-time

Additional typing packages.

No response