I've gotten my RN app to run at the same performance level (as far as the user is concerned) as my native app and there are definitely some non-trivial animations and gesture responders for the RN version.
It did take some tuning and I had to add a couple artificial delays to wait for animations to finish, but nothing crazy.
No opinion on Android yet, but RN for iOS is definitely not negatively impacting the user experience in my case.
It's not in the store yet, but to give you an idea of the UI complexity level - see Slack and Google Chrome's UI and gestures.
It's possible that the OPs ran into certain issues with specific cases or weren't able to figure out a workaround even though solutions exist, but the result of their comments unintentionally create FUD.
I'll give an example. One of the most popular libraries for messenger apps to use in RN is GiftedMessenger, but it's actually doing the keyboard wrong, so people may get the misconception that ReactNative doesn't support smooth keyboard animations, which is not true. There are a lot of examples like this, not meaning to pick on anything specific.
I will say that the library ecosystem around RN doesn't feel mature yet, kind of like Node.js in the early days, but React-Native itself is not the issue.
It drains more battery, it leaves less CPU for other operations, etc.
No, the user always feels. Even the grandmas and grandpas, that are far away from technology as possible, feel it in the back of their mind. They know something is "off".
Also, adding "artificial delays" to wait for animation end is a sign of a terrible terrible animation system.
It's hard to describe without showing you in person and asking you to tell me which app is a RN app vs native, but it's actually the case... There's pretty granular control over the easing functions and the default ones do match the iOS curves.
I was a designer before being an engineer, and I was immersed in Obj-C for 4 years before trying React Native. On the well-implemented React Native apps, it really truly is indistinguishable.
It does help to read obj-c code in the RN repo to see what's possible and how things work so you use the right javascript parameters, but you don't normally have to do additional obj-c coding most of the time.
You're saying that that it is easy to use RN to build an app that is indistinguishable from an app written in obj-c
I'm offering the possibility that your 4 years of obj-c experience trained you to make intuitive judgements about how things should be done and that this intuition is mostly correct. Also that a person without that experience wouldn't make those same judgements and might default to doing things in a lower-fidelity way.
This suggestion is offered without experience using either react native, obj-c, or swift. So coming from ignorance, I might be entirely wrong.
If it's below a certain ms (milliseconds) threshold the user does not "feel". There's nothing more or less to it besides simple physiology.
Additionally I like the React lifecycle better than Android's finite state machine where you always have to know where in the cycle you are, what's next and what's before, etc.
If the wrapper doesn't do what you want, you can still write that portion in native or write your own wrapper with an underlying implementation that does exactly what you want.
So I think RN is a reasonable option for some kind of business that needs a decent, usable app on both iOS and Android. But if your business is mobile-first then I would still recommend you take the time to do a real native implementation.