Trigger cleanup session if taped child process did not exit with success code

! For more details check ops issue https://gitlab.cern.ch/cta/operations/-/issues/994#note_6451613.

---

Problem

• The main cta-taped process launches child processes to handle different tape sessions.
• In DriveHandler.cpp:673, it checks if any of the child process exited abruptly (by not calling exit() or return from main).
  • If it did not exit correctly, it sets the PreviousSession::Crashed flag, which will cause the next forked child process to run a cleaner session before starting.

int rc = ::waitpid(m_pid, &processStatus, WNOHANG);
[...]
if (rc) {
  [...]
  if (WIFEXITED(processStatus)) {
    [...]
  } else {
    [...]
    resetToDefault(PreviousSession::Crashed);
    [...]
  }
}

• However, all exceptions thrown by the Tape Daemon process will be caught in TapeDaemon.cpp:57, and replaced by a return 1 that will be returned by main graciously.
• As a result, the parent cta-taped process will never set the PreviousSession::Crashed flag and the next tape session will not run the cleaner session.

Solution

[Alternative #1] (Simpler and probably better):

• After finding that a child process exited correctly, check if the return code is not 0 with WEXITSTATUS(processStatus).
• If the return value was not 0 (zero), set the PreviousSession::Crashed flag.
  • By looking into TapeDaemon::main(), we know that a child process will only return a non-zero value if an unexpected exception caused it to fail.
• This can be quickly done simply by adding a few lines of code inside if (WIFEXITED(processStatus)) { ... }.

[Alternative #2]

• Simply let the unhandled exception to be propagated up and cause the process to fail.
• This will cause WIFEXITED(processStatus) to return false, properly handle the failed child process and set the PreviousSession::Crashed flag.
• It is slightly more complicated and changes the way how tape server processes may terminate. The end result is the same as [Alternative #1].