These "features" are why I hate Twitter. If Twitter was a personal 90s/2000s shoutbox, I might actually use it.

Without media it could be run at scale cheaper

What will happen is that people will just post links to imgur/giphy etc like they used to when Twitter came out.

Eventually for convenience they’d start getting embedded on the page.

The tracking implications with embedding media would occur. The media would either be proxied (same/more? bandwidth usage, less storage needed though) or just self hosted and then we’re back to Twitter again.

Twitter added features for a reason

You assume that they'd support embedding even though they say their MO is to be text-only? I could see them just resisting support for that, and the type of content adapting accordingly