Fix a problem that access-tab-front-page wrongly caching old page, for example,
even if a user switching to use 'https' from 'http', old 'http' links are
remaining at the front page.

That was because the key of the CachingMethod which was used for caching the
page was ('user' and 'language').
Now the caching key is ('user', 'language' and 'site_url').

Fewer parameters are better if it is not mandatory (Loose Coupling principle).

Revert "Add a test to make sure that even if source and destination are not defined on a mvt, invoice lines are still built"

This reverts commit cf80166b.

Revert "Modify zsql method so that movements without source or destination can be selected in stock table"

This reverts commit 197cf943.

Revert "Modify zsql method so that all mvts are selected in stock table even if they don't have source_section or destination_section"

This reverts commit 87bc3ad5.

 How it happened while upgrading PathTemplateItem:
   1. Try to remove the existing broken object and succeed removing
   2. However it is a broken object so self.unindexCatalog() is
      never called and the old catalog is remaining
   3. Add new object of PathTemplateItem and try to reindex the object
   4. Since old catalog is remaining, the catalog uid and object uid is
      different so catalogObjectList fails with FATAL error.

 The test is tests/BusinessTemplate.py#test_UpgradeBrokenObject.

Acquisition Portal Types is same as for resource.

Example of categories:

journal
  accounting
    consolidation
    national
  project
    deliverable
    r_and_d
  qa
    invalid
    fixed
    wontfix
  trade
    prediction
    reality

Modify zsql method so that all mvts are selected in stock table even if they don't have source_section or destination_section

Add a test to make sure that even if source and destination are not defined on a mvt, invoice lines are still built

It should be consistent with Category setting, because both are used to
render domain tree to user by calling getCompactTranslatedTitle .

We only only need an usual test result proxy, we only need the additional
method getRunningTestCase

This commit in order to force the flush of files
containing test informations after each iteration.

Add option to specify several path where to find users file(s),
and benchmark suite file(s).

This runner is used by scalability test, it communicates
with ERP5 Master and launch test suite (using performance_tester).

Testnode code has been splitted. Runner class has been created for
each test types (unit and scalability).
Communication with SlapOS and ERP5 Master has been added.

These benchmarks are used during scalability tests.
A file which contains list of scalability users is also added.

Contain objects and scripts needed to perform scalability tests.

Without sort the item list is hard to select the variations and most of the time are many

- columns are now searchable
- title display line title if any; for consistency with all other reports
- title is no longer sortable properly. This is a little regression.
- use queries instead of raw where expression

Allow to close a draft position.

trade model and invoice transaction line should be in same invoice