Detect blocked access for users and offer work-arounds

When browsing the web it is very common to receive a blocked access page, or non-resolving "human verification", like captchas. This can be frustrating and I image may push a user to use a different browser.

I wonder if we should start trying to detect some of these cases and offer the user practical work-arounds:

1. Request a new tor circuit. This only sometimes works on the first try, so we could automatically request up to ~10 circuits until it works. Complicated by redirects: #33292.
2. Try and open the webpage in archive.org instead. Usually good enough if you want a read-only page, although sometimes even the archived version is a "you have been blocked" page.
3. Also ask the user to report the site? As recommend in #40086. I'm not sure about this myself.
4. Other ideas?

We should maybe also be upfront to the user that even these work-arounds can fail.

There are also some site-specific work-arounds. E.g. I know that ebay.com will be blocked, but ebay.de is ok, haha. But I think this kind of information would be better maintained by a wiki or similar, rather than our browser.

We could also sync the presentation of this advice with the HTTPS-only error page. #41555 (moved) and tpo/ux/design#238.

Properly detecting cases will likely get messy, so we could trial this behind a preference for a while. I think we should be concerned with false positives (detecting blocking when none occurred) than false negatives (not detecting real blocking). I think if we let the users know about these work-arounds for common sites, that may be good enough to teach them how to handle cases that we don't detect by themselves.

/cc @brizental @boklm @clairehurst @dan @donuts @felicia @jwilde @pierov @ma1 @morgan