CI: Dead link checker

[cwhite911]

I've noticed we have a decent amount of dead links throughout the documentation. I suggest we first do a purge/update of all dead links and then add a new CI check in Github actions to check for dead links in the future.

Form a quick Google I've found the follow tool which seems to be widely used.

• https://github.com/lycheeverse/lychee-action

@echoix what is your opinion?

[echoix]

I've already used it before, for example at the start of #1236. If you need help running it, I think that I stored the exact command line with all bells and whistles that made it work (before md conversion, as this review of html links helped that migration). I could try to dig it out.

That linter was added to
Megalinter (that I also maintain) quite a while ago, when it was still new and less known than now (but still known enough).

I don't think it's appropriate to always run, but might be useful as periodic checkups, either locally or maybe CI. The thing is, we will probably end up with more rate-limiting errors than we would hope for (especially if locally and requesting a lot of GitHub sites for example).

Having jobs fail only because an external server is flaky, slow, certificate out of date or other non permanent issue kinda gets annoying. Yes, it could be possible to properly set up to persist its cache, and limit the number of times the fetches are done (like having it valid for a week or so).