If user omit format means he asked for original data
Guess format from mimetypes_registry later (if any) only to add attachment inside odf archive

progresively slower with data grow.

Add worklist for blocked user login.

Failures.
Add a new report (part of worklist) to show blocked user account list.

This is a list, not a string, take the first occurrence as arbitrary candidate

Clear the Browser history (storing (request, response)) after each
suite. Moreover, it does make sense to keep it as suites are independent of
each other.

It is required to mimic real instance environment

modify the test to follow the change of 6f3992d5 and 5ac7ee53.

imp.load_source does not compile dependencies

Before, if the first iteration failed, the whole execution failed even if the
number of maximum errors specified on the command line argument was not
reached, which was an issue especially when running with a range of users.

The results are now adjusted dynamically after the first successful iteration
rather than at the first iteration.

It should be set according the number of repeats and benchmark suites if the
number of errors is greater to report errors properly to the control process.

Using the same IP address may produce unrealistic results, thus allow to set
the source IP address (only meaningful when not using SlapOS). Network alias
interfaces must be set up before running the script.

Required for mechanize tests, but kept disabled for the moment.

use Permissions.AccessContentsInformation instead of Permissions.View for searchFolder and countFolder, same as ones in ERP5/ERP5Site.py.

As repository can be either bt5 either folder, we must check for both type of id