Due to the new Commentary request forms, which were
also used during vetting (choice=modify), the vetting
went wrong sometimes.
If the vetting action was 'modify', the user was redirected
to the old form which didn't exist anymore.

As a result of how the contruction of this modify structure, the Commentary
was first deleted and thus the data was lost. The requester did receive
a mail that his/her Commentary was accessible on SciPost however.

I fixed this by recontructing the vetting views. I combined the
vetting list and the vet request. I removed the action removing
the Commentary and made this a redirect to a new view which is
a modify page accessible only for EdCol Admins to modify+accept
the Commentary. Also, the mail for 'modify' vetting is also just sent
at *after* modification now.

Add tests for RequestPublishedArticleForm

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

New Commentary requests were not completed
 as the `requested_by` field was not properly
 saved.

Add more VetCommentary tests for the
three different possible valid form actions.

Emails sent from the vet request commentaries
are contains in templates now in the templates folder.

Logic from vet_commentary_request has been moved
to the VetCommentaryForm form. Further moved some
code to improve DRY.

Method test_form_with_duplicate_*'s second part
 gave a false-positive result as the form_data
 contained neither a DOI nor a Arxiv ID. The
 tests came back positive but actually they were
 not.

Move few constants to a seperate file away from models.
This implies minor edits in other apps relying on
these few constants.