Certain actions, such as looking into the editorial page, were
 possible to do by (possible) authors. All submission views
 related to the EdCol are now filtered by the get_pool filter in
 the SubmissionManager to exclude objects with the requesting user
 being a (possible) author.

This commit contains the removal of html from 'the important' models
 or at least contains replacements for model methods such as `print_as...()`,
 etc. etc. Except for `header_as_li`, methods are not completely deleted yet.

> Fixed/Improved the nested comments and comments in reports. They were
 not shown before.

> Improved DRY principle in several templates/codes.

> Created Managers in the `submissions` app to filter i.e. "pool" submissions
 (so exclude potential author from judging on his own publication) for the
 pool view in the query; not in the template!

> More...

Fix bug that allows a forged post request to create a comment,
while the submission is not open for commenting.

and browse) to a Class Based View, using
the ListView class as a base.
Also adds in the permission checks in the
submission process, both in the
PrefillUsingIdentifier as in the
SubmissionCreateView. This should prevent
any unauthorized submissions.

The Comment model now has a option to
 append a file (pdf, png, jpg) to itself.

Start transforming the prefill with Arxiv commentary request
 to a class based view using the Arxiv caller.

Alter ArxivCaller to inheriting from BaseCaller
 and slightly alter the caller workflow. This
 prepares to be able to use ArxivCaller in
 for example Commentaries as well. Further
 initiate empty DOICaller which should
 be used to do similar work but with a DOI.

Prepare ArxivCaller to be used in multiple models.
 It includes the creation of a ArxivCallerMixin in the
 behavior file which should be used in all objects using
 the ArxivCaller.

Move ArxivCaller to the scipost folder, since this caller
 will be used by more apps, such as the commentaries
 apps.

The submission process contains some logic that
could be either handled using default values in the
model (where possible) or abstracted into model
methods. This has the great advantage that the
individual parts can be unit-tested (which is a lot
harder to do with code in the view).

For now there is still some commented-out code for
easy reference. This code will be removed as it is
replaced by the classes above it.