Onion bootstrap server is down: add (a) new onion bootstrap server(s) #1695

According to @g1itch in a comment on #1538, the sole onion bootstrap server is down. This poses extra problems when 'onionservicesonly=true' (old style) or 'onlynet=onion' (new style, see #1694): either bootstrapping stalls or we must connect to a non-onion endpoint.

I see at least two options here:

1. Project maintainers run a suitable node and use that as the onion bootstrapping service. That may actually be the case for quzwelsuziwqgpt2.onion and it is just not properly maintained; I don't know who is responsible for that node.
2. A snapshot is taken and periodically updated of all or some large proportion of onion nodes known to the developers. This snapshot is used in startBootstrappers() in src/network/connectionpool.py.

I prefer the second option as it is much more robust and removes single points of failure/surveillance. However, many of those onion nodes are ephemeral onion addresses and so will quickly be useless. Additionally, some users who may not want the increased load that comes from being a bootstrap node will be included without giving consent.

I am of course open to other ideas, too.