If you like you can use this method to monitor over a longer period of time, and have reusable data.
(Also for the other CPU usage issues you have reported.)
I find this a bit more usable than watching htop. (I don’t doubt the observations mind you!)