Im not really sure what you are comparing here.

* You predownload the "image" which is the part that takes long in docker * You also run inside of `lima` which is already a virtualized layer

What is the thing that is faster? I do love microvms and firecracker, but this article makes no sense to me.