knowledge_graph.py: check outbound links

[m4rc1e]

Some caveats I need to mention:

• I'm only checking for 404 response codes. Many servers will send 403s, 406s etc when using Python requests but are ok when using a webbrowser. If we want to check all 400 codes, I'll need to add more urls to the whitelist.
• We're checking ~500 urls so running the knowledge graph now takes 2 minutes. It used to take < 1 sec. If we want a speed up, we can async the code. Personally, 2 minutes seems fine since we rarely update knowledge items.
• Since it's slow, it's off by default. Include cli arg --check_outbound_links in order for it to run.
• SSL warnings do not fail the CI since these pages do exist.

Fixes #7277

[m4rc1e]

Results I get are:

INVALID url 'https://fonts.google.com/specimen/Source+Serif+Pro?sort=popularity&category=Serif' returned response status code '404'
INVALID url 'https://fonts.google.com/specimen/Source+Serif+Pro?sort=date&category=Serif' returned response status code '404'
INVALID url 'https://fonts.google.com/specimen/Source+Sans+Pro' returned response status code '404'
INVALID url 'https://fonts.google.com/specimen/Source+Serif+Pro' returned response status code '404'
INVALID url 'http://www.typografi.org/justering/gut_hz/gutenberghzenglish.html' returned response status code '404'
INVALID SSL 'https://www.very-able-fonts.com'
INVALID SSL 'https://kupferschrift.de/cms/2012/03/on-classifications/'

[rsheeter]

Nit's but generally LGTM

Edit; ofc please don't submit until CI passes, either by adding a known-failure list or by fixing bad urls.

[davelab6]

by fixing bad urls.

All fixed in #7291

[m4rc1e]

Thanks both of you. Rod, I've implemented as much as your feedback as possible. Dave, thanks for fixing the links!