generate bridge configuration for arti

See merge request tpo/core/chutney!15

The README has a `.md` extension, but was not actually written in
Markdown format, which was causing it to render poorly on GitHub and
GitLab.

because sbws is now using chutney and it would be better the
files are maintained as part of chutney.
This will also help to reproduce #40013 .

Closes #40015

Signed-off-by: David Goulet <dgoulet@torproject.org>

Each chutney network now generates an arti.toml file in its nodes
directory that can be used to connect to that network using arti.

Tor commits 1588767e and 4382e977 broke Chutney and we fix them here.
See tpo/core/tor#40204 .

For #40012

Without these, "None" is put in the ORPort for IPv6.

Signed-off-by: David Goulet <dgoulet@torproject.org>

Let relay.tmpl be IPv4 only else we are forced to pass "ipv6_addr=" to
all templates. Without it, all IPv6 ORPorts end up with "None".

Instead, when we explicitly want a network with IPv6 support, use
relay-v6.tmpl which sets the IPv6 ORPort properly.

Fixese #40008

Signed-off-by: David Goulet <dgoulet@torproject.org>

Our exit-v6.tmpl should always have a v6 ORPort.

Furthermore, bridges+ipv6-min network configuration now uses v6
authorities and a v6 Exit. Without this, the IPv6 bridge self
rechability test would never work.

Closes #40007

Signed-off-by: David Goulet <dgoulet@torproject.org>

Without this check, bridges+hs-v23 seems to be failing.  If this
works with all our supported versions, I'll figure out what to do
with the mentioned issues (chutney#33851 and tor#33852, and also see
tor#33407).

First, bootstrap the authorities and relays.  Next, the bridge and
hidden services.  Finally, the bridge client.

Missing artefact from the pathlib migration.

Signed-off-by: David Goulet <dgoulet@torproject.org>

David Goulet authored Jul 20, 2020 and Nick Mathewson committed Jul 20, 2020

Before this fix, "verify.py" was listed as the chutney tests and failed
to match against the "verify" command passed to chutney.

    (Edited by Nick, to use pathlib instead of os.path)

I seem to have glanced over Path.symlink_to() in pathlib documentation,
or forgotten to change this line of code, one of the two

seems easier to read

Modify both mkdir_p() function signature, return value, and calling
spots to accomodate for the change from os.path to pathlib