If the X server were running in the VM, how would it display things, or get keyboard and mouse input? You’d need to provide emulated hardware devices, which could be something like /dev/fb0 for output and /dev/input/ devices for keyboard and mouse. All that is a very much larger undertaking, and probably a good deal less useful in general. The likes of X are a much more suitable abstraction.

The change here is implementing sendmsg and recvmsg: socket communications, implying communicating with something outside the VM, which in the context of GUI stuff almost certainly means an X server from outside.