cluster-launch-installer-e2e: Restore EXIT and TERM signal handlers

I'd removed these in 3f2f01c
(ci-operator/config/openshift/installer/master: Move to
openshift-install, 2018-09-26, #1677) while removing the shell from
the setup container.  But we got a shell back in b20d571
(cluster-launch-installer-e2e: Start setting expirationDate again,
2018-09-29, #1761), and restoring these handlers will help:

* Trigger early pod wrap-up when the setup container dies.  Both the
  test and teardown containers are monitoring the 'exit' file so they
  can bail early.

* Forward TERM to the installer, so we can exit gracefully when
  someone TERMs the shell script.

For the signal handling to work, we need to keep the shell process
around, so I've dropped the 'exec' from the openshift-install
invocation and moved it to the background instead (so 'jobs -p' will
list it).

ci-operator/templates/openshift/installer/cluster-launch-installer-e2e.yaml

@@ -239,8 +239,12 @@ objects:
       - -c
       - |
         #!/bin/sh
+        trap 'rc=$?; if test "${rc}" -ne 0; then touch /tmp/exit; fi; exit "${rc}"' EXIT
+        trap 'CHILDREN=$(jobs -p); if test -n "${CHILDREN}"; then kill ${CHILDREN}; fi' TERM
+
         export _CI_ONLY_STAY_AWAY_OPENSHIFT_INSTALL_AWS_USER_TAGS="{\"expirationDate\": \"$(date -d '4 hours' --iso=minutes --utc)\"}"
-        exec /bin/openshift-install --dir=/tmp/artifacts/installer --log-level=debug cluster
+        /bin/openshift-install --dir=/tmp/artifacts/installer --log-level=debug cluster &
+        wait
 
     # Performs cleanup of all created resources
     - name: teardown