It's hard to reliable reboot a system over ssh. A common workflow is to
call reboot in a ssh session and then wait until you're able to
reconnect over ssh. Unfortunately, it's sometimes possible to reconnect
before the system has actually rebooted. For that reason, we need an
identifier which is different on each boot to check if the system was
already rebooted or not. E.g. ansible is using the boottime as
identifier [1]. Unfortunately, FreeBSD uses an unintuitive definition
of the bootime. It's not a fixed value, it's adjusted from time to time
[2]. In contrast to FreeBSD, Linux provides a unique boot identifier
called boot_id. Add a similar identifier to sysctl. We're using the same
format for the boot id to make it reusable it in linuxkpi.
[1] https://github.com/ansible/ansible/blob/6ebefaceb6cd0d4961776a94d63a71fc1fc28bc0/lib/ansible/plugins/action/reboot.py#L53
[2] https://cgit.freebsd.org/src/commit/?id=a512d0ab009eedf2f1876fce86d6386bfee626d8