F = pbeta(x,a,b)
Input x matrix (elements between 0 and 1)
a,b positive reals, parameters of the beta distribution
(x,a,b can be scalar or matrix with common size)
Output F for each element of x, F=Prob(X<x) where X is a random
variable with beta density :
x--> x.^(a-1) .* (1-x).^(b-1) ./ beta(a,b)1_{0<x<1}