I don’t think that making Numpy a part of the core Python would do the language any favor. Part of Python’s strength is that it’s relatively small while it still includes a useful standard library.
This modular approach and the ease of using tools like pip for extending it when you need it in a virtual environment is much preferable to having a monolith of a library included for every instance of Python you fire up.
This also makes it easier to develop specialized Python distributions for things like embedded programming or indeed data science, like it's being done today.
People these days seem to think that everyone who uses Python uses it for data science related programming, while that’s not the case by any stretch of the imagination.
Python has its largest chunk of users in the corporate, web and embedded categories.