Time to drop my thoughts on this.
Personally I have no issues with maintaining the forum core + bundled extensions via text files and the Flarum locale commits.
I'm a minimalist, and having everything run through a web of GitHub bots and third party software, instead of a trusty editor only feels overcomplicated to me.
I understand that it would allow for people to contribute updates much more easily in case I'm slacking (which I kind of did on beta 16, sorry about that), but then again, how many new strings are expected to be added in the future? Updating the core is usually a mostly painless process for me with some tricky translations here and there (in which case Weblate wouldn't help me anyways). I'm opting for keeping it simple here and reducing dependencies.
I do however think that using Weblate would make it much easier for people to contribute and update third party translations. I'm not going to keep up with all of them and simplifying this makes alot of sense, especially if the english defaults are already in Weblate and people could just log in and add some translations. I would also be willing to go over translations done by others.
I also think it would be a good idea to have a standardized composer language install command like
composer require flarum-lang/languagenamehere.
This seems problematic with my attempts to leave the core translations in their current state though. Maybe have a repo that merges changes to the language packs into one? I could probably set that up if GitHub actions makes this easy.
I certainly don't want to block any of the efforts made to ensure good and swiftly updated translations in the future though. If me being a snowflake with the German Core translation causes issues I would certainly help on the Weblate front.