DOC: typo in docstring numpy.random.beta, shape parameters must be positive not non-negative

[guilgautier]

Shape parameters of $\Beta(a,b)$ must be positive not non-negative.

See https://en.wikipedia.org/wiki/Beta_distribution

[mattip]

Does the code accept 0?

[guilgautier]

My PR might not be clear enough, sorry for that @mattip.
Let me explain a bit further

Does the code accept 0?

No it doesn't.

That's why there is a mismatch between the docs saying

Alpha/Beta, non-negative i.e. >= 0

And the fact that when passing a,b=0 raises ValueError: a, b <= 0

a, b = 0, 1
np.random.beta(a, b)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-21-97f175c35f60> in <module>
      1 a, b = 0, 1
----> 2 np.random.beta(a, b)

mtrand.pyx in mtrand.RandomState.beta()

ValueError: a <= 0

a, b = 1, 0
np.random.beta(a, b)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-22-76de2fb817e8> in <module>
      1 a, b = 1, 0
----> 2 np.random.beta(a, b)

mtrand.pyx in mtrand.RandomState.beta()

ValueError: b <= 0

Hence my tiny PR to fix

Alpha/Beta, non-negative i.e. >= 0
to
Alpha/Beta, positive (> 0)

[mattip]

Thanks @guilgautier