Explain numpy version requirement

[yannrichet]

version 1.17 seems too old:

$ python3
Python 3.8.10 (default, Jun  2 2021, 10:49:15) 
[GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pylibkriging as lk
>>> import numpy as np
>>> X = [0.0, 0.2, 0.5, 0.8, 1.0]
>>> f = lambda x: (1 - 1 / 2 * (np.sin(12 * x) / (1 + x) + 2 * np.cos(7 * x) * x ** 5 + 0.7))
>>> y = [f(xi) for xi in X]
>>> 
>>> import pylibkriging as lk
>>> k_py = lk.Kriging(y, X, "gauss")
RuntimeError: module compiled against API version 0xe but this version of numpy is 0xd

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
SystemError: <built-in method __init__ of PyCapsule object at 0x7fb0c53586f0> returned a result with an error set
>>> print(k_py.describeModel())
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'k_py' is not defined
>>> 
$ pip3 show numpy
Name: numpy
Version: 1.17.4
Summary: NumPy is the fundamental package for array computing with Python.
Home-page: https://www.numpy.org
Author: Travis E. Oliphant et al.
Author-email: None
License: BSD
Location: /usr/lib/python3/dist-packages
Requires: 
Required-by: scikit-learn, pylibkriging, Funz, scipy, PyWavelets, opencv-python, matplotlib

[hpwxf]

It could be better if we decrease numpy version constraints. To do so, libKriging should be compiled with a defined numpy version (not latest). To install a precise version of numpy, one can do something like:

pip3 install numpy==1.20