Setup translations

[castrojo]

https://v1.docusaurus.io/docs/en/translation#crowdin

We should just set this up ^^^

[lazypower]

I took a look into enabling the i18n of the docs this evening. Thanks for the head start with the sample branch - what copilot came up with looked good to me as a first pass and was functional in the docsy container.

If we focus on going language by language - I think languages can be enabled once or twice by hand, and then it should be nearly automated based on the outcome of that work.

I put together some really rough notes on the process that I'm still developing, but wanted to put this out there if folks are familiar with this process and want to give pointers:

Helpful Links:

• Docusaurus i18n
• Weblate Integration
• Weblate Markdown Errata
• Weblate Continuous Localization

Overly Simplified First Pass at the Process

• Start by enabling the i18n support in docusaurus (this was done in the linked PR)
• Scaffold out the files for translation (Also done in the linked PR)
• Run the docsy container by modifying the docker-compose.yaml to put your local instance in the correct language.
  • command: bash -c "npm i . && npm run start -- --locale es"
• verify the index looks good, all component translations are present.

If you get this far - the basics for i81n are done for docusaurus.

Weblate integration

Here's where we'll need a project owner to make some decisions, and setup the project in weblate.

Can you get in touch with them for the Gratis hosting?

I'm presuming it would be an organization and co-contributors like myself can be invited to facilitate the integration. There will be a few secrets that need to be setup repository side for the build process - a weblate API token that we can feed to the Weblate Command Line (WCL) container for fetching the translations at build time.

It's a little unclear to me if we'll be getting translated markdowns like the crowdin docs suggested, or if we'll be getting back PO|POT files that will need to be transposed into markdown.

In either case - the crowdin integration path was pretty well documented, and I see some similarities here, so this seems doable with a bit of effort.