Proxy path reconstructed from the full URL instead of captured by the route

[abkfenris]

The re_path has no ^ anchor and doesn't capture the proxied path, which forces:

path = request.get_full_path().split("proxy/")[1]

That takes the wrong segment for any URL containing proxy/ twice. Capture (?P<path>.*) in the route and take it as a view argument — the SSRF guards below it (views.py:308-329) get easier to reason about when the input isn't reverse-engineered from the URL.

Sites: views.py:304, urls.py:22

(from #1780)