sap_ha_pacemaker_cluster: Enhance handling of cluster properties and improve ASCS ERS post steps

[marcelmamula]

Reason for changes

1. Post steps for ASCS/ERS were not restarting correct cluster resources, resulting in empty output of HAGetFailoverConfig because Sapstartsrv was not loaded with cluster connector.
2. Cluster properties for stonith were not correct and ASCS parameter was missing.

Changes

• ASCS/ERS Post tasks
  • Reorganize existing tasks (@ja9fuchs This will cause some mess with git diff, so please check it directly in fork)
  • Replace restart Sapstartsrv with restart of resource group so it can manage all resources
  • Add blocks around tasks with same parameters to simplify code
• Cluster properties
  • Remove default variable sap_ha_pacemaker_cluster_cluster_properties to enable proper loading of user defined versus predefined.
  • Remove defaults for concurrent-fencing because it is used only by HANA Scaleout and Azure HA (separate tasks).
  • Fix jinja for priority-fencing-delay which was incorrectly pointing towards previous structure of __sap_ha_pacemaker_cluster_stonith_default
  • Add OS specific __sap_ha_pacemaker_cluster_cluster_properties_default and platform specific __sap_ha_pacemaker_cluster_cluster_properties_platform definition of cluster properties. All stonith-timeout values are defined based on documentations we could find.
• ASCS SAPInstance meta property priority: 100 was added to align with priority-fencing-delay

Tested

Changes were tested on:

• SLES_SAP 15 SP6 on AWS and Azure
• SLES_SAP 16 RC2 on Harvester cluster

NOTE: We have agreed with @ja9fuchs that load order for ha_cluster variables have to be changed in future, to ensure that sap_ha_pacemaker_cluster variables take precedence over them, but this PR was done in alignment with current state (Although comments were added for future reference).

[ja9fuchs]

Looks good.
Tested successfully on RHEL 9.6.