Add relative scaling of variables in CMA-ES optimisation

**Is your feature request related to a problem? Please describe.**
At the moment, CMA-ES minimiser uses a single value of `sigma0` which determines the initial spread of the parameter values around the starting point. However, if the parameters change on a different scale, a single value of `sigma0` will not be optimal for all of them.

**Describe the solution you'd like**
Introduce relative scaling of parameters, as suggested [in the CMA-ES documentation
](https://cma-es.github.io/apidocs-pycma/cma.evolution_strategy.CMAEvolutionStrategy.html#sigma0)

**Describe alternatives you've considered**
N/A

**Additional context**
CMA-ES changes the relative scaling of parameters between iterations, so solving this problem should not change the final result of the refinement.