add 1D minimizer over a [a,b] range

[wkotlarski]

Currently we only have a wrapper over a multidimensional minimizer that uses a gradient descent. This is not the most efficient thing to use if all you want is to minimize a 1d function. For this we should have a wrapper over gsl_min_fminimizer_set. This functionality will be needed both in PRs #446 and #524.

Idea for an implementation: create an overload for the

int minimize(const Vector_t&);

function that works only in 1D and takes

int minimize(double, double);