[krbdev.mit.edu #1622] kshd hanging tests

[Ken Raeburn via RT]

In a few nightly test runs recently on my Linux box (recently updated
to Athena 9.2, based on Red Hat 9), kshd has been hanging.  The expect
process sits there, waiting for it to die, so the test suite
(tests/dejagnu) is still running a few days later.

(1) kshd should probably be dying; if I catch it in this state again
    I'll investigate further what it's doing.

(2) The test suite shouldn't wait forever for it to die; if it doesn't
    die fairly quickly, kill it with -9, maybe wait a few seconds to
    clean up the zombie process, and move on whether or not it died.

Both the krb5-current and krb5-1.3 tests were having this problem.

I don't know if it relates at all to the SIGCHLD/SIG_IGN/wait problem.

Ken