Function minimize_gd 

pub fn minimize_gd(
    f: &dyn Fn(&[f64]) -> f64,
    grad: &dyn Fn(&[f64]) -> Vec<f64>,
    x0: &[f64],
    lr: f64,
    max_iter: usize,
    tol: f64,
) -> OptResult

Gradient descent with fixed learning rate.

Minimizes f starting from x0 by iterating x_{k+1} = x_k - lr * grad(x_k).