In 805374b4 I established the way to run wendelin.core test.  There
component/wendelin.core/test-common.cfg preamble mimics what is setup
by software/neoppod/software-common.cfg in parts for NEO. However I
copied only neoppod into parts without neoppod-develop, and this results
in NEO being installed from PyPI instead of from git checkout.

Let's fix this because everything else - e.g. ERP5 and Wendelin software
releases - use NEO from git checkout which today diverged significantly
from what was last released to PyPI in April of 2019.

Thanks to @jerome for helping to find this out.

See merge request nexedi/slapos!1074

- it's not used in a SlapOS system (seems only used for webcam support)
- it breaks the concept of SlapOs containerization
- finally it tries to create a DBUS file in /etc/dbus-1/system.d/ which is NOT
  allowed (rightfully) for a slapsoft user

This allow caucase test use the random partition IP
address instead of fixed localhost:8000. Which can avoid
the conflict if one test node running multiple test suite.

 - use the requests.Session in a context manager so that resources are
properly closed.
 - close the stdout of caucase process we want to keep silent

See merge request nexedi/slapos!1075

This is used by caucase.sh test

 - use the requests.Session in a context manager so that resources are
properly closed.
 - close the stdout of caucase process we want to keep silent

%Tt was not good, because it also included the idle time from previous
request when the connection was reused, causing times recorded in log
files to be much longer the actual request duration.

%Ta - which is described as total active time for the HTTP request,
between the moment the proxy received the first byte of the request
header and the emission of the last byte of the response body - is much
more suitable for our use case of measuring requests durations and is
similar to apache's %D

Since nexedi/slapos.core!329 we can run use a python3 slapos.core

See merge request nexedi/slapos!1066

This reverts commit d7721eb4.

The change didn't fix the real problem.

There is no reason to wait before banging the partition if VNC is not
available, as reprocessing is the only way to fix the situation.

See merge request nexedi/slapos!1071

This allows to synchronise server position of the mouse with the client one,
thus making mouse really usable.

See https://wiki.gentoo.org/wiki/QEMU/Options#USB

Now that "the bug in the way promises are run" referenced in the comment
in ${template:test} is being fixed, we can start running ERP5 with
python3 slapos.

This was using master branch and when it got renamed to main this
stopped working and we realized that the revision was not pinned.

Also fix a few typos and type annotations

KVM which defaults to 1024 RAM, 10 GB disk and 1 CPU is too small for nowadays
OS requirements, so to avoid confusions, increase the defaults.

Because of a bug in checkSoftware ( nexedi/slapos.core!339 ) we did not check some .so for missing rpath.

Add the missing rpaths now that these are correctly detected by checkSoftware

See merge request nexedi/slapos!1064

Sometimes on testnodes hugo refuses to start because:

    Error: add site dependencies: create deps: failed to create file caches from configuration: mkdir /tmp/hugo_cache/default: permission denied
    Total in 0 ms

new cmmi recipe prevent from having twice the same environment variable.
Also correctly setup swig libraries and fix gobject-introspection compilation.

https://github.com/eclipse-theia/theia/blob/v1.18.0/CHANGELOG.md#v1180---9302021

See merge request nexedi/slapos!1061

Adapt all promise sections to changes in plugin promise recipe.

As Debian Bullseye has been released, set it as first option.

As Debian Buster 10.11 update has been released, upgrade it.

The real predicate to find the master partition is rejected slave publisher
configuration.

After upgrade to OpensSSH 8.8p1 the simple ssh-rsa keys are not accepted
anymore.