Create more homogeneous distributions

[llrs]

At the moment the optimized value is the absolute mean difference with the original value. It could be interesting to fit the entire distribution, which would be a normal distribution 0 centered and sd = ?.

Things to consider:

• the entropy is bounded.
• the chisq.test is also bounded and should follow a uniform distribution close to 0?

[llrs]

ks.test to compare the difference between the distributions, but then it might be easier to test without absolute differences

[llrs]

It might be worth to look at min and max values for each distribution, or at least to quantiles