I have a really basic doubt, which part creates all these directories, where is this configuration of creating directories stored while installing a new OS?
What is the order in which these directories are created? Are they created after /boot has been mounted by the kernel, by which part of the kernel?
The root filesystem is mounted first (readonly) at bootime, as your bootlog could tell you :
[kernel] [ 2.242830] VFS: Mounted root (ext4 filesystem) readonly on device 8:24.
It will then be remounted at init time :
[kernel] [ 2.266181] Run /sbin/init as init process
...
[kernel] [ 6.882156] EXT4-fs (sdb8): re-mounted. Opts: (null)
[kernel] [ 6.882160] ext4 filesystem being remounted at /
From there incidentally populating with (at the very least) all standard base directories (/bin, /usr, /dev, /etc…) possibly empty, possibly serving as mountpoints on which, still as part of the init process…
other disk filesystems will be mounted following /etc/fstab directives :
[kernel] [ 7.155270] EXT4-fs (sdb9): mounted filesystem with writeback data mode. Opts: data=writeback,commit=120
[kernel] [ 7.155274] ext4 filesystem being mounted at /var
[kernel] [ 7.222851] EXT4-fs (sdb10): mounted filesystem with writeback data mode. Opts: data=writeback
[kernel] [ 7.222855] ext4 filesystem being mounted at /home
This as a successful result of some init service (e.g. localmount in openrc), incidentally populating every standard directory of any *x sytem as well as possibly… even more.
BTW, since the kernel does not need /boot filesystem it might well never be mounted. It just can be, at init time, depending on /etc/fstab entries.
Network File Systems can (v.g. not necessarily) be mounted as well at init time, in a way depending on the init system (e.g. netmount service in openrc) taking care network services will have been successfully launched before as an obvious prerequisite.
Special file systems (procfs / sysfs / configfs or other virtual? pseudo? filesystems presenting non-file elements of an operating system as files) will, depending on appropriate CONFIG_* kernel tuneables be mounted at init time, depending on the init system (e.g. execution of /lib/rc/sh/init.sh for openrc) and being consequently automomagically populated.
Temporary File Systems (data is stored in volatile memory instead of a persistent storage device) can also be mounted at init time following directives found in /etc/fstab e.g.
tmpfs /tmp tmpfs rw,nosuid,noatime,nodev,size=4G,mode=1777
/dev particular case :
The /dev directory exists as per / mount together with two nodes the kernel needs at early boot time : /dev/console and /dev/null.
The kernel will add new nodes into /dev directory at boot time following devices' enumeration.
If the kernel is configured to enable temporary file systems, the nodes won't be created in the root partition :
The kernel will first mount an empty temporary file system on the /dev directory, overlaying its content.
[kernel] [ 2.262470] devtmpfs: mounted
Hence allowing the devices to be created dynamically into that filesystem as they are detected (or accessed).
Of course, Linux being a plug&play OS, the population is subject to change at runtime.