Repository (and CMS/workflow) calls session refreshes too often causing high load on database servers (patch attached)

Each call to session.refresh results in query on the database in a clustered setup. We should only make the call if it is really necessary and when there is no lightweight alternative. The clusternode itself already checks the database every 5 seconds for changes which should be often enough for most scenarios.

An easy target to lower the number of refreshes is to make the LoadInitializationModule not call refresh every second. The attached patch seems to have no impact on normal operations.

Other areas where the number of call might be reduced are the workflow with the cms related code and the quartz scheduler.