Empty result of lazy component doubles next node

[aleksandrjet]

I found a bug with empty result of lazy component. I could't fix it and opened pr with fallen test.

Perhaps this is not so easy to fix. Maybe add a notice to the documentation, that lazy component requires something to return (not null or empty fragment)?

[JoviDeCroock]

Hmm, yeah this would be indeed quite hard to fix as it seems a general issue with having no matching DOM for a given VNode during a suspended hydration, basically we want to match some kind of DOM to the suspended node while hydration is ongoing. We currently don't really have a way to see that the oldDom pointer has been consumed and that the future result of lazy will be a placeholder i.e. null.

I'll fold this issue in with #4442 as that would holistically fix these