unfurl: URL corruption and duplication issues (and fix suggestions)

Here’s this comment I made: https://github.com/tox-dev/tox/issues/869#issuecomment-408639666

So it looks like this bot needs improvements in terms of processing inaccessible URLs and outputting duplications.

First two links pointed to different places within GitHub accessible to a logged-in user only. That’s why they’ve got shrunk to just github.com, while my intention was to point to exact pages, which were valid and working URLs.

To fix this, it might make sense to introduce some whitelist of websites, for which bot would preserve original URLs.

Another side-effect of this is duplication of the same resulting URL-preview block in the bottom: we’ve got invalid 2xgithub.com cards and one correct repository card.