When looking for the most up-to-date link to grab a scihub url
we used to use wikiless but now simply directly scrape from
wikipedia.
Functionality is not changed but should work much faster and
more reliably.
Since we now use dotter we can simplify the dir structure for
qutebrowser a lot. Everything dot-filed earlier can now reside
in simple directories called config (for ~/.config/qutebrowser),
data (for ~/.local/share/qutebrowser), and scripts
(for ~/.local/bin) files.