Cross platform differences

[swotlycto]

Hi,

I'm noticing differences between humanized lists created on Windows and on Unix.

The following test passes on a Windows machine:

using FluentAssertions;
using Humanizer;
using Xunit;

namespace Swotly.Core.UnitTests
{
    public class HumanizerTests
    {
        [Fact]
        public void Lists_have_trailing_comma()
        {
            new[] { "one", "two", "three" }.Humanize("and").Should().Be("one, two, and three");
        }
    }
}

But it fails within a GitHub Action running a ubuntu image with:

Expected string to be "one, two, and three" with a length of 19, but "one, two and three" has a length of 18, differs near " an" (index 8).

Initially, I thought this might be a locale issue. But I still get the error, even after setting the locale as follows:

- name: Set locale to en-GB run: | sudo locale-gen en_GB.UTF-8 sudo update-locale LANG=en_GB.UTF-8

I appreciate this is more likely to be PEBCAK than an issue with Humanizer but wondered if anyone else has seen similar