If we're considering switching to TypeScript (which means we'll be performing a near-full codebase refactor anyway), we're essentially crawling our way to a frontend rewrite, and we might as well formally kick off T208256. So there's a bit of a crossroads here: enable transpilation and use new code features for us to rewrite easier later, or we just bite the bullet and focus all resources on a Vue rewrite. Either way, we will need transpilation to some extent, but having to reconfigure from whatever we do here to a Vue setup (at some point in the future) would just create extra work.

I'd be cautious about introducing a build step or TypeScript without first-class support for it in our ecosystem (T199004). One reason that this extension has been difficult to maintain is that its libraries (i.e. backbone + underscore) and setup is different from most other Wikimedia deployed extensions. Keeping things mainstream would be a safer route for maintainability. To that end, I agree that T208256 makes sense, though as I said in that task, I also think it makes sense to use Codex wherever possible, again for the purposes of standardization (in code + UX) and using what is the current best practice.