stack/erp5: make haproxy reload script wait for old worker processes

This is mostly to fix some flaky tests, because with the previous
approach of just sending a signal for haproxy to reload the config
asynchronously we don't know when haproxy has reloaded.
17 jobs for master in 0 seconds
Status Job ID Name Coverage
  External
passed Cloudooo.UnitTest-Master

00:07:42

passed ERP5.StandaloneDeploymentScriptTest-Debian.Buster

00:38:40

failed ERP5.StandaloneDeploymentScriptTest-Debian.Buster.Master

15:02:09

passed SlapOS.Eggs.UnitTest-Master.Python2

00:23:54

passed SlapOS.Eggs.UnitTest-Master.Python3

00:24:09

passed SlapOS.Eggs.UnitTest-Master.Python3Next

00:24:59

failed SlapOS.KVM.ResilienceTest-Master

14:50:58

failed SlapOS.SoftwareReleases.IntegrationTest-Master

08:49:07

passed SlapOS.Theia.ResilienceTest-Master-ERP5

07:24:38

passed SlapOS.Theia.ResilienceTest-Master-Gitlab

17:03:59

passed Wendelin.StandaloneDeploymentScriptTest-Debian.Buster

00:38:29

failed Zodbtools.UnitTest-Master.ZODB6.Python3

00:00:41

failed SlapOS.KVM.ResilienceTest-Master

00:42:41

failed SlapOS.KVM.ResilienceTest-Master

00:37:18

failed SlapOS.KVM.ResilienceTest-Master

00:36:37

failed SlapOS.KVM.ResilienceTest-Master

00:36:21

failed SlapOS.KVM.ResilienceTest-Master

01:01:12