Performance comparisons with Angular and React aren't that interesting anymore. They are good because they got Facebook/Google behind them and not because they are blazing the rest of the Frameworks away.
Right; a large part of the "is this framework worth using" question really relies on "can we get other developers to work on this", "can we make a business case for it", "how is hiring going to work", and critically: "will stack overflow work for us". 50ms on the client is very negotiable and trivial versus cultural concerns.
Mess yourself with Mithril, Mercury or Elm.