The case for the /usr merge

I too have thought of something like this in the back of my head. I was more thinking about device and protocol support and not including "legacy" items (which would require defining what that meant) from the kernel and userland (sort of what you were suggesting).

For example the kernel build would not include support for things like SATA (replaced by NVMe), old school bios (UEFI only), etc. Even going so far as removing IPv4 support in kernel and userland in favor of IPv6 only everywhere.

Clearly this would be a largely academic project, and would probably only boot on specialized hardware or in a VM, but I think it would provide an interesting testbed for certain projects.