This would require an HTML image upload service, which is out of scope for IRCv3 protocol specs.

But nothing stops a server implementation from providing this, and as already said several client+bouncer combinations already support media uploads very well.

The slow moving isn’t the problem of the IRCv3 specs, the issue is the adoption by the large networks and subsequently the clients (which rarely implement features the vast majority of their users on the large networks can’t use).