Reduce server rescue/unrescue times

Server rescue/unrescue is time consuming,
test_rescued_vm_associate_dissociate_floating_ip [23.838659s]
test_rescued_vm_add_remove_security_group [21.670585s]

This is to:
1. create a rescued server in resource_setup to be shared
   between two testcases
2. delete server in RESCUE status directly without unrescue it
   in resource_cleanup
3. move setup of floating ip and security groups from resource_setup
   to inside the testcases because they are used only once

Change-Id: Ic47dc1fa1bcb6ae7f43a6f37565971c94e5494af
1 file changed