Remove / refactor BSFormatter (tracking issue)

[bodymindarts]

While making myself familiar with the code and thinking about improvements I kept seeing complexities connected to the BSFormatter class.

The code base is quite tangled up and that class is one of the places that contributes because:

• it is bucket class with many many unrelated functions
• it has a huge public interface even though most functions don't depend on instance state
• it depends on globals
• it is depended on by many classes
• it is part of a type hierarchy

This led me to refactor it out of existence. The final state of the code is here but it is too much to review and merge in 1 step.
To get the code changes in I can break it down into 5-6 trivial commits / PRs, but don't want to do this work unless other ppl agree that removing this class makes sense.
Please give 👍 or 👎 as feedback on wether I should go ahead. Perhaps I am the only one who has seen these issues, or they aren't hurting anyone and the code can stay as is. In that case I don't want to make the effort.

[bodymindarts]

If you do think this should be taken care of please review the first (trivial) PR #3239.

[bodymindarts]

Reposting the design principles that led me to this refactoring:

• Minimize dependencies on all levels. ie. small public interfaces for classes and libraries.
• Prefer static functions for stateless helpers that don't contain domain logic
• Prefer composition over inheritance for code reuse
• Prefer interfaces over inheritance for polymorphism
• Don't depend on global state in regular classes, better to inject it via constructor

I assume these are not contentious and am omitting justification. Happy to discuss these points if there is dissent.

[bodymindarts]

Here is step 2. Extract 2 classes #3244

[bodymindarts]

Completed