n3fit - Separate the PDF from the rest of the model

Created by: scarlehoff

This is a purely technical change (from the outside nothing should have changed... yet) to decouple the PDF from the rest of the model. The reasons for doing this are basically three:

1. Remove a lot of hacks that were in*
2. Use the PDF as a standalone thing outside of n3fit: now it is easier getting rid of the writer.py so that we don't reinvent the wheel (see for instance #558 (closed))
3. In my mind it is a very necessary step for optimization, GPU running, etc.

When doing this change I added a few TODOs here and there but I prefer to do them as separate PR because a) like that they are kept small b) I see as different issues.

Let me polish it a bit. Feel free to volunteer yourselves as reviewers, otherwise I'll volunteer people once I finish.

*this is something I should have done a long time ago, but at the time was easier to keep compounding things on top of the model