l10n runbot builds are all failing when running at least one test depending of AccountTestCommon because it:
- doesn't create a sandboxed testing environnement to manage the multi-currency, multi-company, the default company's currency, the exchange rates...
- doesn't setup a testing user then all tests are done using the superuser.
- doesn't provide a fully setup chart of accounts: exchange difference journal is not set, accounts have bad types, etc...
- is run sometimes at-install.
--task: 2301180m
Raphael Collet