This is the downside to relying on Storybook's internal APIs... non-breaking updates can break how axe-storybook-testing works.

I still think it's worth it, though. It's the best, easiest, and most performant way I could come up with to actually render and iterate through all the stories in a storybook.

So that they do not ignore legit typescript errors.

Wait until a story has truly "completed" rendering, which we can see at https://github.com/storybookjs/storybook/blob/06bf39e29b517650b303c1f4fa4fbf2697ad3ce1/lib/preview-web/src/PreviewWeb.tsx#L604.

The main use case for this is waiting for the new "play" functions to be finished before starting the axe checks.