I've actually come up with a solid rule of thumb for when it makes sense to use: If you have 2 async streams of events or data, that's at least 4 success scenarios you have to code for:
1. A has emitted, B hasn't
2. B has emitted, A hasn't
3. A and B haven't emitted
4. A and B have both emitted
Now at 2 sources (and ignoring the error cases) that's right on the edge of being able to mentally model it and not forget a case, but at 3 you have 9 cases, at 4 you have 16, etc. etc.
The branches and code you'd be creating to deal with this situation procedurally would be far more unreadable compared to the RxJS syntax. It was always a complex problem.
Regarding Angular and RxJS, I don't think Angular really embraces RxJS (other than maybe the Async-Pipe). Actually I think it often feels awkward to work with RxJS in Angular, especially when you use observables that are not used via the Async-Pipe, unsubscribing for instance is annoying at best.
I agree with this sentiment, we've been waiting patiently for observable inputs for a ridiculously long time: https://github.com/angular/angular/issues/5689, but I do have a pretty solid pattern for dealing with subscriptions these days:
1. Create one subscription as a private member of your class: https://github.com/spaceribs/plopdown/blob/4823f5017733ee785...
2. In your OnInit, subscribe and add that to your main component subscription: https://github.com/spaceribs/plopdown/blob/4823f5017733ee785...
3. Unsubscribe from everything in your destroy method: https://github.com/spaceribs/plopdown/blob/4823f5017733ee785...
I've been really happy using RxJS outside of Angular. I know the author isn't claiming this is the case, but Angular adopting RxJS is a double-edged sword in that it seems many people equate RxJS with Angular. RxJS + React Hooks is a really powerful pairing to mitigate the hooks madness of more complex components. I use RxJS for the data flows and hooks just to bind the observable to a DOM rerender.