contentValues() nowadays return also non ERP5-ish objects such as python
scripts , OFS.File etc. Because of that, if an ERP5 Folder contained a
non ERP5 object, which does not have fixConsistency or checkConsistency
methods, Folder.checkConsistency was causing an infinite loop because it
was acquiring checkConsistency method from self.

Reduce the time required for a user to scan a multiple pages PDF.

It has been fond that the slowest part of the scanner dialog was to wait for an image to be uploaded before being able to scan another image.
In order to improve this, it has been decided to allow user scanning a document even if the previous image is not yet uploaded.

See merge request !1034

Having WebPage_viewAsWeb in default skin selection break other sites
relying on default behavior of erp5_web.

See merge request !1071

erp5_document_scanner: Rename web_page_module content IDs to not conflict with erp5_web_renderjs_ui path pattern.

- Stop relying of CSS hide/show attributes. Instead, only put needed elements in the DOM.

- Upload images asynchronously to improve the usability

Add a tales expression as default field value. The default value is a JSON containing 2 keys:
* the relative_url of an active process
* the list of active process image content to aggregate
By default, when the field is rendered for the first time, this tales expression is evaluated and a new active process is created at that point.
So, default json is like: {active_process: 'portal_activities/1234', image_list: []}
* store the active_process relative_url in the gadget state in the render method
* store the image_list in the gadget state in the render_method
* change getContent method to send a json build from the active_process and image_list state

- Change the gadget checkValidity method to return false when:
  * no thumbnail has been generated
  * a thumbnail upload is in failed state

- POST image cropped to backend
  * change the `extra renderjs` parameters list to pass a Python script to call when a new image is cropped.
  * change the gadget to use jio.util.ajax with the script full url (provided by the server)

- Translate and Update i18n info
- Ajax must return a active process image content UUID
- Add spinner in button to show that we are saving it in background
- Add reference to Active Process. With this, we can search easily active processes used by document scanner and remove them
- Add an alarm which will delete outdated (>5 days old) active_process created by the dialog.
- Improve code to switch cameras in mobiles with more than 2 cameras
- Display intermediate image after get orientation and instantiate Cropper

Also, move test_data folder to Products.ERP5.tests as this does not need to be in the ZODB.

And we should keep erp5_core as small as possible.

Display like this:

![image](/uploads/0e182e76a94c56eb450e8ef354a9c5c1/image.png)

Where we can easily read in listbox the history of item movements: 
```
   → A
A  → B
B  → C
```

instead of

![image](/uploads/24bea37097dc22c2eaedf323d049aeb3/image.png)


```
A 🠐 
B 🠐 A
C 🠐 B
```

( as always, this also enable coding style test )

/reviewed-on nexedi/erp5!1061

This was failing [coding style test](https://nexedijs.erp5.net/#/test_result_module/20200305-AB828AA2/54)

/reviewed-on nexedi/erp5!1070

Keep compatibility with existing officejs tests, by keeping the
data-i18n attributes, even if useless.

Chrome will deprecate appcache in the coming weeks.
As appcache has been dropped on Firefox, this change will improve the speed on Firefox.

The service worker only provide a cache for all ERP5JS gadgets.
ERP5JS will still be usable on browser without service worker API.

The list of files to cache is directly added inside the service worker
code, in order to be able to skip the usage of the fetch method (which
does not yet cover all browser HTTP usage).

This list is configurable on the web site level.
No need to change the service worker code directly.

The service worker will be updated as soon as the web site modification date changes.

End user does not need to reload its browser tab.
Only browsing ERP5JS will trigger the service worker update.

When a new update is found, ERP5JS will automatically reload itself if
there is no risk of losing user data input.

An empty appcache is kept, to allow migration of existing client to the
worker automatically.

See nexedi/erp5@2e3f7a8a

Use the precache script to get the list of gadgets.
Ensure to also check the launcher gadget.

Report subgadget service error.

I don't know how it was exported, but this object does not install:

     File "/parts/erp5/product/ERP5/Document/BusinessTemplate.py", line 4141, in install
       return ObjectTemplateItem.install(self, context, **kw)
     File "/parts/erp5/product/ERP5/Document/BusinessTemplate.py", line 1422, in install
       container._setObject(object_id, obj)
     File "/parts/erp5/product/ERP5Type/Base.py", line 253, in __call__
       result = self.__dict__['_m'](instance, *args, **kw)
     File "/parts/erp5/product/ERP5Type/Core/Folder.py", line 1055, in _setObject
       return folder._setObject(self, *args, **kw)
     File "/eggs/Products.BTreeFolder2-2.13.5-py2.7.egg/Products/BTreeFolder2/BTreeFolder2.py", line 443, in _setObject
       self._setOb(id, ob)
     File "/parts/erp5/product/ERP5Type/Core/Folder.py", line 970, in _setOb
       return folder._setOb(self, id, object)
     File "/eggs/Products.BTreeFolder2-2.13.5-py2.7.egg/Products/BTreeFolder2/BTreeFolder2.py", line 256, in _setOb
       ids = mti.get(meta_type, None)
   AttributeError: 'NoneType' object has no attribute 'get'

This portal type refered a non existant init script, which prevents
adding documents of this type.

Having WebPage_viewAsWeb in default skin selection break other sites
relying on default behavior of erp5_web.

Replace it by domsugar.

This speed up gadget class generation (handlebars template parsing is
slow).

This attribute is only needed for the translateHTML method.
There is no reason for it to be kept in the final DOM HTML.

Add service worker precache script.

Add service worker precache script

renderJS now correctly handles relative URL.