1 minute is too agressive for general use.

  The upgrade ordering seems buggy and requires to force reinstall it

This allow during a subscription process, create a user and them create all
documents using that user w/o rely on proxy roles (keeping good ownership
across the documents).

Without this (on slapos master ui) the redirection places value unquoted causing parser to fail later on when resolve {n.me}

  The user can be present from diferent logins, password tool only works with ERP5LoginUserMager
  users.

Mixed commit:

  43b1ae1b
  2f05fe1b

Returning true allows to add movements to existing one, and this group shall
always separate them.

Note: Generic code shall have no constraints at all.

Conflicts:

	bt5/erp5_accounting/bt/revision

If 2 lines have the same effective date (catalog has a 1 second precision),
always return the validate and open first.

Commit: 02d06501

Changes applied from history:

http://git.erp5.org/gitweb/erp5.git/history/refs/heads/interaction-drop:/product/ERP5/Document/SubscriptionItem.py?js=1

Original commit from : Lukasz Nowak <luke@nexedi.com>
                       3e45ec35

Also minimise activity hurricane by calling expand directly.

Open Orders are searched using indexation timestamp, which allows to see them
in "windows", and does not repeat expand if not needed.

Generate activities and allow to pass the tag. Use search and activate
everywhere which will allow to walk through objects without killing the cluster
even in case of really big documents.

Avoid calling isDivergent which can take few minutes to finish.

causality_state comes from well designed causality workflow, which informs
enough about delivery state. Fetching causality_state property is extremely
fast.

Notes:

Do nothing on 'calculate' instead of disabling *_causality_interaction_workflow as
these interaction workflows do not generate activities directly, and do work
that can't be postponed.

By contrast, 'calculate' transition of delivery_causality_workflow is modified
so that no 'updateCausalityState' activity is created by default.

This commit also update list of method_id which should not call calculate_causality

Conflicts:
	bt5/erp5_crm/bt/revision

See SQLBase._getExecutableMessageSet for operation principle.
Removes the notion of order_validation_text: activity validation is no
longer evaluated per-activity , but per-dependency for multiple activities
at a time. In this context, order_validation_text does not make sense as
it flattens all dependency types for a given activity.
Rework activity-dependency-to-SQL methods: use a dict rather
dynamically-generated method names.
Based on initial work by Julien Muchembled.

Seen happen on SQLDict with 94 indexation activities being stuck assigned
to a processing node on which load failed because of an SQL deadlock.

The main reason for not using cache.addAll was to fill the cache as a low
priority task, not to generate too much traffic, the safari compatibility
was secondary. Now that safari seems to support cache.addAll, remove the
comment about safari.

Components also have some extra checks in checkConsistency, we should
include this in coding style test.

See merge request !1324

This simplify code and also bring the functionnality of c484f8aa
(erp5_xhtml_style Base_cancel and logged_in: do not allow redirection outside
ERP5 site., 2016-02-12) for the cases where authentication_policy is
installed.

This also fixes a problem with translations of "Your password will expire at
 {date}", which was using different messages for every possible date.

Tests needed to be updated because we now redirect with properly URL encoded
parameters.

See nexedi/erp5@ea219b74

When new password does not match policy, in reset password and change
password dialogs, we used u' '.join([str(message) ...]) to join all
translated messages in a string, but this construct will decode the
str(message) to unicode using ascii, so it will fail when these messages
contain some multi bytes characters.

Extend test coverage to check that these dialogs uses translations and use
non ascii messages in the tests, to make sure we don't have regressions
with this issue.

See merge request !1001