When you set the extent scene for a Text Object I noticed there's an asynchronous delay between the time you change the text and the time the extent scene screen transform gets adjusted. It doesn't happen in real time.
I noticed this because I wanted to show several text objects one after the other vertically and the text can have multiple lines (speech bubbles). So what I have to do is wait 1 second before I make any changes to the extent scene object after I change the text. It seems that even waiting 100 ms works but I'm not sure if slower devices require more time.
Ideally I believe the extent scene should have it's frame adjusted in real time so we can make changes to it after setting the text. In my case I need to measure the height so I can position speech bubbles vertically one after another.