This should fix the problem that `Revision History` is unusable on test
results because tests have a different name when they have been running
on different paths

Although these are synonyms here now, it may evolve in the future.

Some underscore-prefixed properties may fail callable() check (ex: because
the document requires some properties set to access an underlying actual
callable).
So give these a chance to be skipped before failing.

[erp5_odt_style] [renderjs_ui] Allow global Print action from ODT style only on Documents and not Modules


-  split Print action into two and set object_jio_print only the one for Documents
-  add all possible actions to activate the "Export" button

/reviewed-on nexedi/erp5!600

erp5_real_time_inventory_accounting: Split the BP into dedicated ones for Sale, Purchase and Internal.

Checking 'delivery' portal_type for PL type will not work on real
implementations as the PL is not built during the first expand and
expand_related=0 after and this is how it should be done anyway.

Same as IPL use case.

IPL must expand both "In"coming (only PPL until now) and "Out"going (only SPL
until now) SMs, so split up ledger='stock' TMPs and add accordingly two
subcategories to stock ("entree" and "sortie").

Also instead of using 'use' categories everywhere (TMPs and Rules), use the
Delivery Portal Type as this makes things clearer for IPL use case.

This is enough for now to handle both Outgoing/Incoming/Transit use cases.

This implements Outgoing/Incoming Accounting Rules and Builders.

This column is not a significant condition for this test. It is an
unreliable transient piece of information intended for a human observer
monitoring activity execution. processing_node is the reliable piece of
information this test should care about.

For when redundant activities may be skipped.

By providing a method id receiving considered document and activate_kw as
parameters, and producing a new activate_kw value.

Julien Muchembled:
- ExceptionRaised class simplicitation

Two almost identical methods, each having two 75% identical code
paths: this screamed "factorisation".

This method must not be used, so do not mention it in an example script.

Caching in authentication-related work is not a good idea.
Also, simplify the resulting script:
- do not load Password Event from ZODB, just use the creation_date found by
  catalog
- reorder accesses to only do them when they are needed (ex: preferences)
- factorise access to persistent objects (portal_preferences)
- fix coding style: space after comma, no space around named-argument equal
  signs, 2-spaces indentation, spaces around multiplication and division
  operators, upper-case pseudo-constant
- factorise now vs. expire_date comparison, getting rid of the mishandled
  "now == expire_date" edge case in original code
- remove commented-out code and obvious comments
- DateTime is already implicitly bound in any script's globals, no need to
  explicitly import it.
In turn, this allows removing immediate reindexation on corresponding scripts,
as immediate reindexation will be disallowed for restricted python.

See nexedi/erp5@8fdc1a15

Add tests to prevent future regression.

This is not a regression, only the test getting better at spotting an
incorrect behaviour which pre-existed.

Also, while exporting catalog methods, explicitly check for `expression_instance` property.
Now we keep expression_instance property in a volatile attribute, so we should prefer using
it.

When data-gadget-renderjs-extra attribute is not set to an element,
`element.getAttribute("data-gadget-renderjs-extra")` returns null.

Possible error messages :
- sub_key is null
- TypeError: Cannot set property 'key' of null

/reviewed-on nexedi/erp5!602

In case of catalog methods where we have expression empty or None, trying to get
expression_instance raises an error. So, as to keep the compatibilty with how it
used to check for value of expression as shown in this diff :
nexedi/erp5@04f331a2
its better to check the same in mixin.

This commit:

- Adds a new Activity called "SQLJoblib"
- Adds a Backend to be used by joblib
- Uses OOBTree to store results instead of ConflictFreeLog
- Adds a getResultDict API to fetch resut Dict

It uses the original work from rafael@nexedi.com and loic.esteve@inria.fr