If that was the reason behind the change, changing the signature to strictly be `_jsx(name, attrs, children)` would've made more sense. I also didn't see this

If that was the reason behind the change, changing the signature to strictly be `_jsx(name, attrs, children)` would've made more sense. I also didn't see this reasoning stated in the RFC thread.