Add kube::util::wait-for-jobs to report status, unlike builtin wait

** Sigh **

Fixes an additional complaint in #6463

hack/lib/util.sh

@@ -95,3 +95,14 @@ kube::util::host_platform() {
   esac
   echo "${host_os}/${host_arch}"
 }
+
+# Wait for background jobs to finish. Return with
+# an error status if any of the jobs failed.
+kube::util::wait-for-jobs() {
+  local fail=0
+  local job
+  for job in $(jobs -p); do
+    wait "${job}" || fail=$((fail + 1))
+  done
+  return ${fail}
+}