contain hello0, hello, decode, encode

change tools.decrpt func to use tempfile instead create a never removed
file

Because some bugs of nemus occur in test-nodes, when use popen occur
"RuntimeError: Error from slave: 500 Unknown command", or when add
veth name option error. I write sample module my_net to mangae the net
ns.

my_net: use unshare to create net namespace, then use nsenter to enter
ns correpand to execute cmd like add device, set ip address... and
cotain some network demo for running tests.

re6st_wrap: wrap the deploy of re6st node, ease the creation of cert
file and run of the node

test_net contain the tests:
  - ping test: net segment(router), multi-net segment(demo),
	       test after a reboot
  - hmac test: test on demo

ping is like ping file in demo, a script to run multiping

registry: add test for RegistryServer and registryClient,
testRegistryServer mainly test methods concerned to http rpc request. Other methods that call request_dump like
getBootStrpPeer is not include.
testRegistryClient test the rpc call with or not with "cn" parameter.

cli.conf: test each situation call the cli conf.

tunnel: add test for BaseTunnelManager, MultGatewayManager

tools is a util cotain method to make cert and ket files

On Debian < 10:
- debhelper extends postint with 'try-restart', which caused re6st to
  remain stopped after upgrading from a re6st package that stopped
  services unconditionally in prerm (see previous commit);
- deb-systemd-invoke does not check whether the service is enabled
  so 'try-restart' can't be replaced blindly with 'restart'
  (as it is the case in recent Debian).

In addition to minimize re6st downtime, this fixes upgrade from
versions that still use init scripts.

When we drop support for Debian 8, we'll increase debian/compat
and we'll revert this commit.

To prepare for the removal of geoip2, we want nodes to ask the registry
for their country. geoip2 is kept in this update since nodes will still
need to figure out countries of other nodes which haven't updated yet.
Once all nodes will be updated to this version, geoip2 will be ready to
be deleted.

See merge request nexedi/re6stnet!32

If a node restarts and has its address with its country in the cache, it
won't set the _country attribute, and will therefore ignore countries
when making tunnels.

See merge request !31

Addresses passed to _updateCountry can contain countries if
cache.my_address is used to initialize address in BaseTunnelManager init
function, which causes resolve to be called with too many arguments.

In some circumstances, the hello_protocol attribute could get modified
on the wrong peer, which would raise an AttributeError.
On reception of seqno 1 from a peer with protocol < 7, protocol could be
equal to zero which would cause handleHello to not return True, causing
the handshake to fail.

This commit concerns networks that use the --same-country option.
We recently discovered that the IP geolocation database contains
incorrect entries. To work around this, the protocol needs to be
changed by adding the country as 4th field in addresses (the first 3
are: ip, port, protocol) and the new --country option allows a node
to announce a country that differs from the one the GeoIP DB.

Thanks to the previous commits it's possible to implement backward
compatibility, by not sending the 4th field (country) to nodes that
can't parse it. Of course, these old nodes would continue to not
create appropriate tunnels and after a while, the administrator of
the network may decide to increase registry's --min-protocol (7).

In a network with only nodes that implement this last version of the
protocol, the nodes may only use the GeoIP DB to resolve their own IPs.

See merge request !27

There is a need to be able to extend the protocol without breaking
compatibility with old nodes. This is done by sending version.protocol
during inter-node handshake, in seqno 1 and seqno 2, so that a node
knows what version the peers speak and use appropriate format.

This is implemented with partial backward compatibility: handshake with
an old node succeeds when the new node does not have to send seqno 1.

Behaviour changed in commit 5a8e4186.

On machines using systemd, services were not enabled by default.

To reenable RTT-based metric, we usually want to force old nodes
to upgrade, so that they don't always look faster that others.

This reenables RTT-based metric, which
was disabled since we moved to v1.8+.

See commit 5a8e4186.

Commit 40d4e496 is not reverted
because Python 3 will also require to keep the distinction between
blob and text.