Prevent generating wrong urls for now

romain_dev: forum

romain_dev: forum use html viewer

romain_dev: forum improve post rendering

romain_dev: thread reader use page language inqtead of local

romain_dev: forum set expected link on the listbox cells

romain_dev: thread fixup

romain_dev: thread use line separator instead of background color

romain_dev: thread paginate

romain_dev: forum link to last page

romain_dev: thread simplify conf from erp5form

romain_dev: forum attachment

romain_dev: thread force white background

This reverts commit e50d05af.

Do not access form submission REQUEST from the listbox list method,
as it is rendered asynchronously in ERP5JS

This reverts commit 35b2c024.

Allow edition in the new UI

This make everything slow as hell and prevent to quickly save.

Example: <h2 class="foo">bar</h2> => <h3 class="foo">bar</h3>

erp5_web_renderjs_ui: keep previous focus color

Change max age to 1 hour and 1 day

Nothing uses redirect_url, so do not bother setting it.

For consistency with what is being allowed in the Restricted monkey-patch.

Such errors are not retried with the retry adapter, so retry them
directly.

Now that default cloudooo is updated to LibreOffice 7, this revealed some
problems with percentage values in spreadsheet cells, it was apparently
working by chance on LibreOffice 5.

This improves erp5_ods_style to use `style:style` referencing a
`number:percentage-style` through their `style:data-style-name` attribute.

Because the same thing was missing for dates to follow the input order of
datetime fields, this is also handled at the same time.

We could do something similar to support all input styles of float fields,
so that we support `1 234.5`, `1.234,5` or `1,234.5`, but this is not done
this time.

See merge request nexedi/erp5!1861

 - the test was using wrong attribute and it was using DummyLocalizer,
which breaks the site configuration. Fix the test and use mock
 - fix ods style not to translate untranslatable columns

Page templates have a builtin translation system (with i18n:) and an
"unicode conflict resolver", so it was not necessary to explicitly
convert to unicode on python2 and this simplifies the transition to
python3