following format defined in system preference level.

so that we will not have HTTP authentication dialog in case of  Unauthorized exception to prevent blocking in functional tests.

Create a sql cache for getInventory call
Remove full-inventory optimisation as it now useless
Implement full-inventory feature directly into stock
Make erp5_pdm depends on this optimisation
Install optimisation for all unit tests

> Note, ImageMagick assumes the sRGB colorspace if the image format does not indicate otherwise.
http://www.imagemagick.org/script/command-line-options.php#colorspace

Also updated expected converted result with sRGB output.

This reverts commit a4072868.
=> it makes test testERP5WebWithDms failing. Please update unit
test and also add test to check what you fixed with this change.

Adjust test accordingly and fix bad test.

> Note, ImageMagick assumes the sRGB colorspace if the image format does not indicate otherwise.
http://www.imagemagick.org/script/command-line-options.php#colorspace

First select pads for current user, then if it's anonymous user and it's preferred so return public knowledge pads for context. This change allows logged in users to have their own customized pads independently from value of a web site preference : "Force Anonymous Gadget " and still we can show or hide public pads for context for anonymous.

In case still we want to show default pads for logged in users we can use preferences which can contain knowledge pads templates. The templates can itself be customized which allows a KM or ERP5 UI site to show a set of default pads for any context which with a single click can be adjusted by user to his needs.

it should not be based on it but on title to check dependencies
if only provider is possible, use it
wrap long lines

Conflicts:
	product/ERP5/bootstrap/erp5_core/bt/revision

This adds support for IPv6 and new syntax for unix sockets.

This deprecates the old syntax, which uselessly requires to specify credentials.

This makes code simpler, faster and easier to understand.
It is easy to forget that ZODB connections can be reused by different threads,
which led to bug such as the one fixed by commit 2c11b76a.

ZODB already maintains a pool of connections to reuse
so we don't need anymore to have one.