Document NumPy 1 interaction with max simulation size

[mhucka]

What is the task?

A user reported being unable to simulate models with more than 32 qubits. It turned out they were using NumPy 1.26, and updating to version 2.x made the problem go away. This is something we should document somewhere.

(Optional) Do you have ideas or preferences for the approach?

No response

How urgent is this for you?

P1 – needed by the next release

[kris524]

Hello, should requirements.txt be updated to require numpy >= 2.0.0?

[mhucka]

@kris524 Thank you for your interest. In the current versions, we want to avoid requiring NumPy 2, to maximize compatibility. This will probably change in the future because too many other things will come to rely on NP 2, but for the time being, that's the strategy.

This particular issue is more about documentation than modifying the requirements.