Fitted parameters shouldn't be globally accessible

[rozyczko]

From the Summer School's retrospective:

During the fitting, the parameters that were defined for the fit are being changedimplicitly, which I find very dangerous. While we had a discussion whether this ismore intuitive for users with a background in e.g. Matlab, I would argue that thelibrary should follow Python's best practices

[damskii9992]

This is an older issue that we've known about for a while.

We have 2 options:

1. We change the interface to the Fitter to make it more pythonic and user-friendly.
2. We make a new public Fitter class and make our current Fitter class an internal class for technique libraries only.

In either case we need to look at existing fitting libraries and decide on the public API we wish to have.

I am more in favour of the 2nd option, as this then won't be a breaking change and we can more easily split responsibilities. Like we know that the public Fitter always receives its parameters in a list whereas the internal Fitter always receives a ModelBase class to extract parameters from etc.