Avoid integration to fail due TCP server timeout

timeout 60 might be too short for someone with lots of network
interfaces.
Since we grab the PID and run the server in the background, the timeout
is not critically necessary.

scripts/runtests.sh

@@ -234,8 +234,7 @@ if [ "${RELEASE%.*}" -ge 20 ]; then
     # Test TCP connectivity (system_setup only)
     clean
     port=50321
-    LANG=C.UTF-8 timeout --foreground 60 \
-        python3 -m system_setup.cmd.server --dry-run --tcp-port=$port &
+    LANG=C.UTF-8 python3 -m system_setup.cmd.server --dry-run --tcp-port=$port &
     subiquity_pid=$!
     next_time=3
     until [ $next_time -eq 0 ] || [ ! -z "$(ss -Hlt sport = $port)" ]; do
@@ -268,7 +267,7 @@ if [ "${RELEASE%.*}" -ge 20 ]; then
             fi
         done
     done
-    kill $subiquity_pid
+    kill $subiquity_pid || true
     if [ $loopback_failed -ne 0 ]; then
         echo "Loopback was expected to connect"
         exit 1