Project 3: Sled db reopen could fail on `could not acquire lock on ....` on Linux

[k0Iry]

Should we add some timeouts after we kill the server process?

let handle = thread::spawn(move || {
    let _ = receiver.recv(); // wait for main thread to finish
    child.kill().expect("server exited before killed");
});

When we kill the child process in which server is running, if open the sled db immediately before LOCK_UN takes effect, we probably will get this error.

Maybe adding some timeouts like 50ms or so in between can make the test cli_access_server_sled_engine more reliable.

[k0Iry]

Also see the issue for sled here: spacejam/sled#1234