It's enough to set on the class and if we don't have it on instances, it
saves a few bytes in pickles

There have been a regression recently

When installing scripts or components from a business template, set
the _erp5_coverage_filename property so that the plugin is able to find
the corresponding code from the business template filesystem
representation.

coverage might use these to compute the upload URL for coverage data

coverage needs to be started before importing the modules, so it can
not be started so late in runUnitTest, it has to be started before
importing any Products/ERP5* code, ie. in slapos wrapper script.

See https://coverage.readthedocs.io/en/6.5.0/faq.html#q-why-do-the-bodies-of-functions-show-as-executed-but-the-def-lines-do-not
for more details about the rationale

Use super to always call the overridden method.

Extend the API to pass the business template being installed.

Call the hooks also when installing sub objects, they were not called
in one place.

Introduce an action script which redirects to the alarm, so that
"something happen" and user can see that the method was invoked. This
also makes it possible to use this action in selenium tests

With ERP5 workflows, transition id is different (it has a prefix),
we must use transition reference

by using getDocumentValueList to lookup all documents at once instead
of using getDocumentValue for each document

test names are easily longer than the 20 characters that was set
in this field

This does not seem to be an actual problem because the object is
declared protected, but these missing security declarations make
testSecurity fail.

We have been using `filter(lambda x:'/erp5/' in x[0], error_list)` as a
way to see only missing declarations from ERP5 code, but only ERP5
filesystem code contains /erp5/ in the filenames, in-ZODB components
filename is set to something like <portal_components/document.erp5.FTPConnector>

Change to also check methods for which filename contains <portal_components
and also to print the ignored methods

The logic to guess groupable lines was considering source of
destination side of an accounting transaction depending on
preferences. This was not good, for two reasons:
 - with internal transactions we want to consider both the source
  and the destination
 - if user preferences are mis-configured, grouping would not work.

switch to a logic where we consider both source and destination
sides to prevent these problems.

python xmlsec's xmlsec.so contains multiple level module in the same
.so (xmlsec, xmlsec.template etc) and this seem to cause problems to
astroid, trying to lint a module containing:

  import xmlsec
  xmlsec.template.encrypted_data_create('...')

cause an error:

   File "develop-eggs/astroid-1.3.8+slapospatched001-py2.7.egg/astroid/raw_building.py", line 360, in _set_proxied
      return _CONST_PROXY[const.value.__class__]
  KeyError: <type 'module'>

and unlike similar error happening with ffi's module from cryptography,
retrying does not help.

This version of pylint tries to import the C modules for analysis and
get confused by support ffi's CompiledFFI type.

Use our utility  _register_module_extender_from_live_module to register
a stub module instead.

This also reverts b21f83fc,
"BusinessTemplate_getPythonSourceCodeMessageList: Retry in case of
pylint internal failure." which is no longer necessary with these
changes.

 - the transaction is aborted in tearDown, no need to abort
  explicitly, for same reason if we want to delete documents, we
  need to explicitly commit (see TestERP5Person_getHateoas_mode_search)
 - simulate decorator should not commit the transaction to keep this
  semantic of abort by default, which keeps test isolated.

There was a regression with 45c03413 (hal_json_style: prevent a
translation of worklist with document count, 2021-05-10), the title
of the worklist was translated twice. This was supposed to be
catched by the test, by checking that the mocked gettext had only
one call for "Draft to Validate", assuming that the translation of
"Draft to Validate" was "Draft to Validate" and that a double
translation would cause this to be translated twice, but because of
a side effect of another test inserting a translation for
"Draft to Validate", this problem was not noticed.

This fixes the double translation and adjust the test to not depend
on the actual content of message catalog.

This is still needed for now, see 19327cc0

because it contains legacy workflow objects.

because it contains legacy workflow objects.

workflow: support old data using legacy WorkflowHistoryList on Python 3 where WITH_LEGACY_WORKFLOW is False.

workflow: now another workflow script should be accessed by container.script_<SCRIPT_ID>, instead of container.<SCRIPT_ID>.