Add NMPreprocessor protocol and Preprocessors class

[toni-neurosc]

This change gives Preprocessor classes the same treatment that Feature classes were being given in my other PR #307 , that is:

• I created a Protocol "NMPreprocessor" from which all preprocessor classes inherit
• I moved the initialization and execution of preprocessing from "nm_run_analysis" module to its own module "nm_preprocessing"
• Like the class Features, I created a class Preprocessors that imports the necessary modules according to settings and instantiates them.
• The NMPreprocessors instances are then run on the data by using Preprocessors.process_data(data)

I also added some utilities that can be shared by nm_features too:

• a NamedTuple class in nm_types called ImportDetails (not sure I like the name a lot) that holds data about the name of a class and the module it's located in
• a function get_class that lazy imports a class and returns the class object, given an ImportDetails input. Right now I put it in nm_types, but i don't know if that's the right place for it

This was quite a difficult modification due to the differences in the arguments taken by the different Preprocessor classes, so it requires quite a bit of "meta-programming" to get it working without errors. But I took some valuable lessons regarding typing and also regarding the different "settings" needs of the different preprocessors.

[toni-neurosc]

Awesome, merging #305 here did indeed fix the testing errors. Merging this into main then.