[Epic] Refactor uncertainty notebook

[redeboer]

It's better to split up the work for #82 into multiple PRs. This issue is to keep track of them.

I originally tried to implement #82 directly, but this resulted in different uncertainties despite using the same 9 models. So to find the cause, it's better to work step-by-step.

• #101 (unrelated issue)
• #102
• #103
• #100
  (will affect the decay rates and weighted polarization values a bit)
• #107
• #109
• Main target of the refactoring:
  #104

Each of these steps (apart from #100) should not affect the fit fractions and their uncertainties or any other quantities computed in the notebook! Only then can #82 be addressed.