Remove the "--disable-static" configure option. Fix the warning message: "configure: WARNING: unrecognized option: '--disable-static'".
See merge request !1231

Resolve `@theia/core` subdependency `vscode-uri` to a locally installed patched version, to workaround https://github.com/microsoft/vscode/issues/154048.

This fixes access to Theia through the IPv6 `backend-url`. That issue was discussed in nexedi/slapos!1204

The patched version of `vscode-uri` is published here: https://github.com/xavth/vscode-uri/tree/fix/ipv6_url_authority

To do this, `yarn.lock` was regenerated, so this also caused a version up to Theia 1.28.0.

See merge request nexedi/slapos!1227

See merge request nexedi/slapos!1217

Instead of lying in tests and checking for NotReadyYet in the connection
parameteres use promise to assert that generation of KeDiFa URLs is really
ready.

Drop hacks applied in the test, so that it will uncover all missing promises
and corner cases.

bootstrap-script-url and data-to-vm are important features of
enable-http-server and are expected to work correctly, thus cover them
with instance-level testing.

In mean time fix detected issues.

See merge request nexedi/slapos!1222

  The content of inline is a jinja2 template

If cluster name is not found, use default cluster name

We disable sql_log_bin while managing repman_slave_definition database and
destroying slave database. Else, changes are replicated which can break
slave node.

edgebot is dropped too, as it was shared instance implementation for edgetest.

Some Py2-SR like neoppod/stress-testing need Qemu but not Ceph.

Also note that the conditional [systemd-markupsafe-download] only works
if [systemd-python] uses the same Python as the one that runs buildout.

rbd functionality is enabled only on machines with supported libc version.

See merge request nexedi/slapos!1209

The mariadb/proxysql service automatically start when repman shut it down then
repman cannot check if service was correcty stopped because it's already up.
Stop a service will not work too.

It should be possible to add 2 or more cluster in the same instance

Rarely used https backend was missing the backend Via additions, thus it was
not transmitted there.