[16.0][l10n_br_fiscal][IMP] l10n_br_fiscal: standardize data+demo load

[rvalyi]

Standardize the loading of all l10n_br_fiscal data + demo files.

In l10n_br_fiscal we have some very specific challenges:

• loading all the production fiscal data for the tax engine takes 3 minutes and a lot of CPU, this is a no go for demo and tests. This is why we had these subset demo files for NCM, NBM, NBS, CEST...
• for mission critical cases, some fiscal data might better be altered by some fiscal power user and keep noupdate=True. If noupdate were False, reloading the l10n_br_fiscal module would also take a lot of time. By default modules data files load with noupdate=False.

This explains the peculiar loading hook we used in l10n_br_fiscal/hooks.py until now.

But after working on the isolation of the tests from the generic demo data, I also got convinced we could finally standardize the loading of the data + demo files.

In fact all it takes is a custom monkey patch of the odoo.tools.convert.convert_csv_file method to force noupdate=True for some files and to be able to filter and load only some demo records in demo/test mode for some large csv files.

Benefits:

• standard loading of data + demo files; simpler
• no more "file-not-used" pylint warnings we could not even filter (in v18 warnings will we block the CI by default)
• removed 800 lines of demo files we had to maintain
• no more production/demo data discrepancies (the remaining tests errors seem due to that!)
• will also help simplify the loading of fiscal fixtures in [16.0][REF][l10n_br_base][l10n_br_fiscal][l10n_br_account] tests don't depend on demo #3563

[OCA-git-bot]

Hi @renatonlima,
some modules you are maintaining are being modified, check this out!