administration: workaround alarm.getLastActiveProcess not during alarm execution

Describe the problem as a code comment in Alarm.getLastActiveProcess for
now

bt5/erp5_administration/SkinTemplateItem/portal_skins/erp5_administration/Alarm_activateCheckConsistency.py

 kw = {}
 
 if context.getProperty('incremental_check'):
-  last_active_process = context.getLastActiveProcess()
+  last_active_process = context.getLastActiveProcess(include_active=True)
   if last_active_process is not None:
     kw['indexation_timestamp'] = {
         'query': last_active_process.getStartDate(),

product/ERP5/Document/Alarm.py

@@ -344,11 +344,19 @@ Alarm Tool Node: %s
   def getLastActiveProcess(self, include_active=False):
     """
     This returns the last active process finished. So it will
-    not returns the current one
+    not returns the current one.  ( XXX this docstring looks wrong. What about include_active ?)
+
+
     """
     if include_active:
       limit = 1
     else:
+      # XXX bug !
+      # This cannot be trusted during alarm execution.
+      # When alarm is running, isActive returns, then we assume that the latest
+      # created active process returned by the catalog is the "currently
+      # active" active process, but if the currently active active process is not yet indexed, we
+      # this method does not return the latest active process, but the previous one.
       limit = self.isActive() and 2 or 1
     active_process_list = self.getPortalObject().portal_catalog(
       portal_type='Active Process', limit=limit,