Because Unix and Linux have a decades old tradition of documenting with
man pages (and, on GNU systems,
info files ...). See man(1), man(7), man-pages(7). BTW,
man command and pages are optional (and you won't install them on every Unix system).
The file system hierarchy is described in hier(7).
It is defined by the Filesystem Hierachy Standard available on https://wiki.linuxfoundation.org/lsb/fhs
Several filesystems, notably
/proc/ (see proc(5)) and
/sys/ (see sysfs(5)) are pseudofile systems provided by the kernel code. You don't want to bloat the kernel with extra code producing such
README-s (which is useless to the vast majority of users). Even the kernel's configuration file is only optionally available as
/proc/config.gz which is often disabled in most kernel configurations. And many Linux systems are embedded systems (e.g. your smartphone, your smart appliance or IoT device, your RaspberryPI) where resources are scare enough to avoid being wasted.
/sys/ is mostly useful to sysadmins and to developers writing low-level utilities, and both are supposed to be able to find documentation appropriately.
Why not to place
README files into the hierarchy to make it easier for people to learn what's going on
If you really want such
READMEs, write your own loadable kernel module providing them, or setup some unionfs to provide them. I don't think it is worth the effort (and a unionfs on
/sys would probably slow down your entire system).
Remember that kernel code consumes RAM (it is never paged out and sits in physical memory, not virtual memory), even if not used. So it makes sense to avoid bloating it.