Normalized scales

[JorisVincent]

Implement normalization of scales, and of confidence intervals to range [0, 1].

Implemented in the wrangling, and an option in the overall scale estimation function.

With tests, also against the example datasets analyzed in R.

Closes #25

[JorisVincent]

Now that I'm thinking about it, is doing this an option the best way? It's very straightforward, cheap calculations -- both for normalizing the scales and the bootstrap. It doesn't require rerunning the estimation or the bootstrapping.

We could also just include both the unnormalized and the normalized scales (and CIs if calculated) in the scale_estimation output dict? What do you think, @guillermoaguilar ?

[guillermoaguilar]

I think it is ok as it is, usually we do want only want of those, either normalized or not. So it is ok as a parameter option.
Looks good, merging!