I do also feel like it is more jittery (android 13, jerboa 0.0.35)
On your own email server make sure that you have correctly setup dkim, spf, reverse dns, and the ip of server is not on any known denylist - but it still doesn't guarante that your email server would not get flagged as a spam source.
Or maybe setup transport rules that would try to deliver most emails directly, but to major providers like Gmail/outlook which are quite picky via smtp relay - at work we are using AWS SES to do that, it is not that expensive, but it depends on the volume.
As you are running cloudflared in docker container, by default 127.0.0.1/localhost equal that docker container. Easy fix is to add
extra_hosts:
- "host.docker.internal:host-gateway"
to your cloudflared service in docker-compose. Then you can point tunnel to host.docker.internal and it should point to your docker host instead of localhost of container. (ofc after reloading services by docker-compose up -d)
Another easy solution would be to make your owncast server listen on lan IP of your server, and point cloudflared tunnel to that
I also sometimes experience that issue, check in developer console if you don't have errors with websocket connections. Not sure what causing that (my side/server side), but most of the time CTRL+F5 (refresh with clearing cache) fixes the issue