ptrace: Add a test using PROC_REAP_KILL to kill a stopped debuggee

[markj]

Summary

This exercises the patch in D44523.

[markj]

Remove some leftover debugging cruft

[kib]

Why do you need the killer child to kill the debuggee?

I suppose the test is run in dedicated process, so the debuggee is automatically detached, continued, and then the zombie is gathered either by init or the test process?

[markj]

In D44564#1016180, @kib wrote:

Why do you need the killer child to kill the debuggee?

It's not strictly needed I think. It just help exercise more code paths in reap_kill_subtree_once().

I suppose the test is run in dedicated process, so the debuggee is automatically detached, continued, and then the zombie is gathered either by init or the test process?

Yes, ATF tests always run in a dedicated process, so we can let init reap zombies.

[markj]

Make the test case useful. reap_kill() will not single-thread an
already-stopped process.