Localization stuff & humanize improvements

[Baptiste]

This patch fixes some strings that xgettext didn't like, and allows the localisation of contrib.humanize. It adds also the option "--from-code UTF-8" to the xgettext command line (in bin/make-messages.py), to solve a little encoding issue.

This patch adds the support of exponents in contrib.humanize.ordinal.

{{ value|ordinal:True }}

will output :

1<sup>st</sup>

By default, no change, but that can be helpful.

[Baptiste <baptiste.goupil@…>]

Oops, there was a little issue

[Malcolm Tredinnick]

The first part is a dupe of #3410, which I am just about to commit. The second part is a wontfix. If you want customisations like that -- and there are so many possibilities that we cannot accommodate everybody, so we take a fairly minimalist line -- it's easy enough to write your own filter.

Please file one issue per ticket in the future. It's makes triaging/resolving much easier.

[Malcolm Tredinnick]

I've ended up taking part of this patch, too, since you caught the plural problem in the humanize filters.

[Malcolm Tredinnick]

(In [4704]) Fixed #3410 -- Edited a few i18n markups for completeness and to remove some
warnings from recent gettext versions. Refs #3704. Thanks, Michael Radziej,
mirrorballu2@… and baptiste.goupil@….