This reverts commit 68d0c366.
There are other codepaths which call _postCopy without going through
__duplicate. For example: manage_renameObject (called on parent by setId).

@nexedi 
This will add a form setting field 'action title', in order to setup the text content of the form action button.
So far the action button text content is the title of the form which can be inconvenient some times. 
If 'action title' is not set-up it fallback to the form title.

![Screenshot_2019-09-04_at_18.50.04](/uploads/e3b49369f347c64a14fa0218c5923c9f/Screenshot_2019-09-04_at_18.50.04.png)

/reviewed-on !925

In testFormDialogActionTitle, check if it's well fallback to form title

See 28774376

By importing the new TimerService unconditionally,
the commit 34c81caf
(Timeout: introduce publisher_timeout configuration)
broke automatic site creation & publisher timeout
when the zope config specifies the old timerserver.

It makes to no sense to synchronize with the clock if there's nothing
after to remain synchronized.

[erp5_ingestion] Do not use absolute URL parameter
This is not compatible with ERP5JS

This is a fix for 5eeb52d5. Waitress
also requires a cancel method. Not having one raises an uncaught
AttributeError.

otherwise min / max criterion does not work.

Fix remaining functional test failures and drop a few more code that become unused now that we have a `setFile` command.

Because firefox 68 executes event handlers for events triggered on disabled elements ( https://phabricator.services.mozilla.com/D10000 ), Zelenium is also adapted to mimic the behaviour we had with firefox 52 (and chrome behavior) of not executing event hander.

TODO:
 * [x] `testContributeTab` sometimes displays Converted, sometimes Converting. I wonder if we can wait for something to make the test deterministic, otherwise we can assert a regexp.
 * [x] `erp5_web_renderjs_ui_test:testFunctionalRJSCore` ( discussed in 05b9c4af (comment 87278) )
 * [x] someone update test nodes to use new firefox.
 * [x] understand why with firefox 58 selenium sends clicks on disabled button.

/reviewed-on !922

Volatile attributes may disappear from one statement to the next.
Make code reliable by moving it one level up the call stack, removing the
need for volatile attributes.
_setId and _postCopy get back to their original "copy of OFS.CopySupport"
status (see comment at beginning of file).

This BT contains the same category twice.

Bank Account, Currency{, Module}, Currency Exchange {Cell,Line} already
exist in erp5_base, which is a dependency of this Business Template at
least through erp5_trade.
"composition" and "variation" Base Categories exist in erp5_pdm, which
is a dependency of this Business Template via erp5_trade.
"site" Base Category exists in erp5_base, which is a dependency of this
Business Template via erp5_trade.

classification is already provided by erp5_dms, which is a dependency
of this BT already.

requirement is also present in erp5_project. So add a dependency on that BT
and remove the duplication.

This BT achieved the rare feat of containing the same document twice.

region/europe/western_europe is also present in erp5_tiosafe_test, which is
already installed in the same unit test. So add a dependency on that BT and
remove the duplication.

currency_module/euro is also present in erp5_tiosafe_test, which is already
installed in the same unit test. So add a dependency on that BT and remove
the duplication.

Vincent Pelletier:
- Disable timeout by default.

Vincent Pelletier:
- Ignore TimerService-originated publication requests.
- Add copyright headers.
- Disable timeout by default.
- Give a meaning to requested negative deadlines.
- Docstrings.
- Assorted simplifications.

The Assigments are only relevant if the User is based on Person Document, on SlapOS, an user come from a Computer or a Software Instance Document. This change makes the document compatible with other kinds of Users, as Assigment is overkill for Non-Person cases, as the presence of ERP5 Login is enough to determinate the User existence. 

The other alternative would be re-factorize to not duplicate code, which seems not the intention of how this code was written. 

/cc @jerome, @kazuhiko 



/reviewed-on !458

Simpler, and reduces the exposure of exc_info (which can be tricky to use
in other contexts).

When we updated Firefox from 52 to 68, the behavior of firefox changed.
With 52, when calling element.dispatchEvent on a disabled element, the
event handlers were not executed, but in 68 they are.

Unlike Medusa, WSGI does not do it.