at this point they all fail after long timeouts, because neo does not
support py3 yet

Not sure why it was OK on python2 and not sure if this has to be done
or the promise code needs to cast the results of getConfig()

Otherwise it fails with the following exception because of non-ASCII characters
(`[A-Z&é@{]{3,7})`) (see unconvert()) below):

  => erp5_core/PathTemplateItem/portal_preferences/default_site_preference.xml
     File "zodbpickle-2.0.0-py3.7-linux-x86_64.egg/zodbpickle/pickle_3.py", line 844, in load
        dispatch[key[0]](self)
     File "zodbpickle-2.0.0-py3.7-linux-x86_64.egg/zodbpickle/pickle_3.py", line 1035, in load_short_binstring
        self.append(self.decode_string(data))
     File "zodbpickle-2.0.0-py3.7-linux-x86_64.egg/zodbpickle/pickle_3.py", line 989, in decode_string
        return value.decode(self.encoding, self.errors)
  UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 19: ordinal not in range(128)

This reverts commit 6c399134.

Directly expose all passlib.hash supported hashes, using a `passwd-`
prefix. For example, to access `sha256_crypt`, use `passwd-sha256-crypt`
option name.

  [secret]
  recipe = slapos.cookbook:generate.password

  [config-file]
  hashed-password = ${secret:passwd-sha256-crypt}

The strategy for compatibility is that:
 - haproxy still listen on the same port as before, without rewrite rule.
   This is called "legacy" port.
 - for each frontend from request parameters, we introduce an haproxy
   frontend with a rewrite for the corresponding `internal-path`
   parameter.
 - the shared frontend instance is updated to use this new frontend
   entry from haproxy. This will cause a small downtime until the shared
   frontend is updated to the new URL on ERP5, but since this feature
   was not used, it's OK.

Technical details are that we:
 - split haproxy config to have frontends and backends.
 - introduce one frontend in haproxy for each frontend from request
   parameters.
 - routing-rule-list argument is still honored the same way, globally
   and after path from frontend.
 - change the shared frontend requests to use "" type, no longer "zope"
   type.
 - we don't do automatic detection of /VirtualHostRoot in URL but always
   add it, because it could be used to trick zope into thinking it
   serves requests for an arbitrary host and do open redirects
 - before using the request's host header in virtualhost path, we check
   that it does not contain /, to prevent injection of virutalhost path
   elements through the host header.
 - we don't use the "path" parameter from shared frontend, because we
   want the frontend to be simple, so we don't want it to rewrite the
   request path (which is also the reason why we deprecated "zope" type)
 - the tests have changed a lot, because they were using what's now the
   "legacy" URL types, so we updated it to use the new URL types with
   all the /VirtualHostRoot/../ in path and also because they use IPv6
   URL, no longer IPv4

and save the already allocated ports in a state file, so that requesting
new families does not change already allocated ports.

This reverts commit 620c9332 (stack/erp5: stop using caucase managed
certificate for balancer, 2020-11-10) with an updated design. We add a
caucase service for balancer in the balancer partition. The caucase
service from the root partition (that was not used) is removed.

The underlying idea is that the default configuration should use multiple
caucases with limited scope, here we have one caucase to manage the
certificate used by haproxy server in the balancer partition, so we put
one caucase to manage this certificate and the caucase is configured to
auto-accept one certificate only. The plan is that when we will add a
certificate for mariadb server, we'll add another caucase inside this
mariadb server.

For more advanced usage and also to support the cases where a new
certificate needs to be re-emitted for some reason, users can request
with an existing caucase URL. In that case, they will have to accept
the certificate requests.

Notable changes:

balancer/ssl/caucase-url is no longer documented in parameters, this is
an internal parameter, users can pass one global caucase service to
manage all partition

CAUCASE environment variable is no longer set when running zope. There
was no identified use case and with this new approach of multiple
caucases, the term "caucase" alone became ambiguous.

This is not documented in schema and has no effect in erp5 (but this is
still used for slapos-master)

This change the format or the (mostly) unused frontend parameter to
support requesting more than one frontend and also enable the request of
a frontend by default, so that requesting a frontend separately is no
longer needed.

The `frontend` parameter now also supports requesting frontends for
specific paths on the ERP5 backend, the example below requests a
frontend serving directly a web site, with the necessary rewrite rules:

```js
{
  "frontend": {
    "default": {
      "internal-path": "/erp5/web_site_module/renderjs_runner/"
    }
  }
}
```

The example below requests a default frontend to the erp5 root, to
access the ZMI or erp5_xhtml_style interface and two web sites:

```js
{
  "frontend": {
    "default": {},
    "erp5js": {
      "internal-path": "/erp5/web_site_module/renderjs_runner/"
    },
    "crm": {
      "internal-path": "/erp5/web_site_module/erp5_officejs_support_request_ui/"
    }
  }
}
```

The example below has an explicit definition of the zope families using
`zope-partition-dict` parameter, because there is more than one zope
family, no frontend is requested by default:

```js
{
  "zope-partition-dict": {
    "backoffice": {
      "family": "backoffice"
    },
    "web": {
      "family": "web"
    },
    "activities": {
      "family": "activities"
    }
  }
}
```

Continuing this example, to have frontends for backoffice and web
families, the frontend request can specify the families, like it is
demonstrated in the example below. In this example, we don't specify an
entry for "activities" family, so no frontend will be requested for
this family.

```js
{
  "frontend": {
    "backoffice": {
      "zope-family": "backoffice"
    },
    "web": {
      "zope-family": "web",
      "internal-path": "/erp5/web_site_module/web_site/"
    }
  }
  "zope-partition-dict": {
    "backoffice": {
      "family": "backoffice"
    },
    "web": {
      "family": "web"
    },
    "activities": {
      "family": "activities"
    }
  }
}
```

our new nexedi.org currently no longer serve static pages with links,
using absolute links should allow us to have test passing until this is
fixed.

These partition references should be kept short, they are a mechanism
to use a short path for unix sockets, because unix socket paths can not
exceed 108 characters.

When running the test in theia, this was causing errors:

   # [ALERT]    (100453) : config : parsing [/srv/slapgrid/slappart15/srv/runner/instance/slappart7/tmp/inst/with-max-rlimit-nofile6/etc/haproxy.cfg:37] : log : socket path '/srv/slapgrid/slappart15/srv/runner/instance/slappart7/tmp/inst/with-max-rlimit-nofile6/var/run/log.sock' too long (max 97

Currently, due to ensure_ascii=True default of json.dumps, we are
insisting on our JSON schemas to be ascii-only and all other characters
to be represented by \uxxxx escapes. So far this was not problematic as
all our schemas contains only ASCII characters, but upcoming
ors-amarisoft changes want to use e.g. "→" symbol:

    https://lab.nexedi.com/kirr/slapos/blob/b51f5523/software/ors-amarisoft/software.cfg.json#L15

which currently results in failure of json-schema test:

    FAIL: test_ors-amarisoft_software_cfg_json_format (slapos.test.test_json_schema.TestJSONSchemaValidation)
    ...

    First differing element 14:
    '      "title": "\\u2192  eNB/gNB | Radio Unit",'
    '      "title": "→  eNB/gNB | Radio Unit",'

And in general, in 2023 I think there is no reason to insist on our schemas to
be ASCII-only: say if one wants to describe something about "α" parameter. It
would be good to use that α character directly and seeing it in the editor,
instead of using escapes all the time.

As indicated by below stackoverflow answer "JSON spec requires UTF-8 support by
decoders": https://stackoverflow.com/a/594881/9456786 , and indeed checking
JSON specification also confirms that by default JSON decoders shall use UTF-8:

    https://datatracker.ietf.org/doc/html/rfc7159#section-8.1

This way, I think, we can switch to UTF-8 safely.

/reviewed-by @jerome, @lu.xu
/reviewed-on nexedi/slapos!1498

See merge request nexedi/slapos!1499

/reviewed-by @kirr and @jerome
/reviewed-on nexedi/slapos!1494

wendelin.core 2.0.alpha3.post9 supports golang 1.21: with this new
version we can set golang 1.21 as the new default.

/reviewed-by @kirr and @jerome
/reviewed-on nexedi/slapos!1494

Go1.21 already has the fifth minor revision (released 2023-12-05) and
should therefore already be sufficiently stable. Furthermore we need it
to fix a bug in a NEO/go dependency [1].

Please find all details in the official release note:

    https://go.dev/doc/go1.21

It was released on 2023-08-08 [2]. Due to the go promise of compatibility
most software should still compile without any problems.

In golang < 1.21 we needed to patch golang to fix
https://github.com/golang/go/issues/42525. In golang 1.21 we still need
to apply a fix, but can't apply the old patch because the code changed.
In golang > 1.21 this problem is already fixed with https://go-review.googlesource.com/c/go/+/520055.

Because of https://github.com/golang/go/commit/092671423cd95eaa6df93eb29442fef41504d097
'TestUnshareMountNameSpace' fails on golang 1.21 [3]. In golang 1.20 this test
was skipped [4]. To fix this failure the additional patch
'skip-unshare-mount-test.patch' has been added.

---

[1] nexedi/wendelin.core!22 (comment 195769)]
[2] https://go.dev/doc/devel/release
[3] --- FAIL: TestUnshareMountNameSpace (0.18s)
    exec_linux_test.go:243: unshare failed: exit status 2
        unshare: mount /tmp/TestUnshareMountNameSpace2210137852/001 failed: 0x1
[4] === RUN TestUnshareMountNameSpace exec_linux_test.go:333: kernel prohibits unshare in unprivileged process, unless using user namespace — SKIP: TestUnshareMountNameSpace (0.00s)

/reviewed-by @kirr and @jerome
/reviewed-on nexedi/slapos!1494

This version is compatible with gcc 12

using https://github.com/ilevkivskyi/com2ann

drop versions not running on debian 12 and add new versions

This was disabled to "keep compatibility with current font selection
problems", but I don't think we need to care about compatibility for
this. This was anyway causing a problem that system fontconfig will be
used if it's available, so it's better to define things explictly to
prevent falling back to system configuration.

This caused cloudooo to select different fonts, because LibreOffice-bin
comes with some Noto fonts and fontconfig now prefers Noto font