WSL2 uses it's own init system - only one kernel (plus a tiny initrd) is virtualized by hyper-v; each subsequent linux instance (distribution) is containerized. There's additional facilities that handle resource allocation dynamically vs the user specifying a static amount during VM creation.

These (and more) result in the end-user interacting with WSL2 the same way they would any normal application. To think of it as simply a VM isn't quite correct.