Function stepsize_blend 

pub fn stepsize_blend(
    old_x: &DVector<f64>,
    new_x: &DVector<f64>,
    max_step: f64,
    factor: f64,
) -> DVector<f64>

Applies step size reduction and capping.

Algorithm

1. Compute displacement: dX = newX - oldX
2. Scale by factor
3. If ||dX|| > max_step, rescale to max_step
4. Return oldX + scaled_dX