[Bug #20995] Protect `IO.popen` block from exiting by exception

...pending

If two threads are running, with one calling waitpid(-1, Process::WNOHANG),
and another calling waitpid($some_pid), and then
$some_other_pid exits, we would expect the waitpid(-1,
Process::WNOHANG) call to retrieve that exit status. However, it
cannot actually do so until $some_pid _also_ exits.

This patch fixes the issue by calling do_waitpid unconditionally in
waitpid_wait; this will ensure that a waitpid -1 actually reaps
something (after first checking that no PID-directed call wants the
process).

[Bug #20490]

	Don't clear pending interrupts in the parent process. (#10365)

If two threads are running, with one calling waitpid(-1), and another
calling waitpid($some_pid), and then $some_other_pid exits, we would
expect the waitpid(-1) call to retrieve that exit status; however, it
cannot actually do so until $some_pid _also_ exits.

This patch fixes the issue by unconditionally checking for pending
process group waits on SIGCHLD, and then allowing pending pid-only waits
to "steal" the notification.

[Fixes #19387]

* Remove `require 'io/wait'` as it's part of core now.

* Update ruby specs using version gates.

* Add note about why it's conditional.

https://github.com/ruby/ruby/actions/runs/3562236318/jobs/5983796689

Check for that the daemon process is detached, that means it is not a
child and not waitable.