Over the years, we included almost all possible default group by
criterions for getMovementHistoryList, so that in essence, we do not
group. A better solution is to just not group by by default, this
achieve the same and we do not have to filter duplicate in SQL (this
filter operation is expensive with large data sets).

Use "Transaction Type" everywhere and stop using "Journal Type".

On accounting transaction views, it is "Transaction Type".

In l10n_fr, l10n_de, l10n_pl both "Journal Type" and "Transaction Type"
where translated to the same term.

This cause slow queries and returning rows that user won't be able to
group because we only allow to group transactions from the account
anyway.

because credential_request.submit() is now guarded by Modify portal content permission.

ProgrammingError could be raised for various raisons, while the code
was assuming that the only raison was missing table. Improve by making
sure the error is about missing table, if not, raise.

Left panel should not be always visible on a laptop screen

expr(md) raises NameError when expr references a non-given argument, so
catch that exception too. (sqlvar catches Exception, which seems overkill,
so don't follow it)

Fix missing import.
Actually follow sqltest render API, which is different from sqlvar's.

mrp simulation: allow to customize transformation rule to define to reverse input source/destination

Somehow two files didn't made it to the recent deploy. I am still to figure out the reason but I suspect it might be due to my recent rebase.

/reviewed-on nexedi/erp5!166

- use mixin class to allow reusing common performance functions
- remove usage of quiet and run_all_test

Reduces the number of meaningless 404 errors.

so that we keep the query string in the redirect url if exists.

Instead of only having possibility to check when inventory will be below a reference_quantity,
allow to check when it will be higher.

This code is useful to run any unit test in a live instance. Until all products tests (like
ERP5/tests/*) are moved as components, this code is useful.

This reverts commit 9240d498.

also stop casting select_language to str.

If ERP5 Jupyter kernel tried to persist an object created from a new-style class, it would crash because the object has no `_p_jar`. This makes the creation of the ObjectWriter fails. This kind of object needs to be handled by (c)Pickle.

Related to bug report https://nexedi.erp5.net/bug_module/20160819-882FA2, created by @kasra.

@Tyagov, please review when possible. 

/reviewed-on nexedi/erp5!162