High: crmd: Also purge fencing results when we enter S_NOT_DC

The prevents us from prematurely erasing a node's status section and
starting resources in more than one location.

Sequence:

• Node A is DC
• Node A is fenced
• Node B becomes DC
• Node A returns
• Node B shuts down
• Node C becomes DC
• Node C incorrectly clears out Node A's status section
• Node C starts services that are already active on Node A
  • Because we tell the PE Node A is safely stopped when it isn't