Incorrect platform status should prevent sync

[alittley]

Problem

• In the current system, platform status doesn't have an affect on whether a node syncs
  • Instead, we rely on gossip being started or stopped, and on the following checks before accepting or initiating a sync: fallen behind, reconnect complete, PCES replay complete
• In the case of PCES replay complete, this approach causes undesirable behavior:
  • If the platform status state machine doesn’t transition out of REPLAYING_EVENTS due to a problem during replay, the node will start syncing regardless, wasting network resources and filling logs with errors about events that could not be added to the hashgraph.
• Since the platform status has no direct affect on whether a node syncs or not, it is possible that an unknown bug could cause a node to sync even when the platform status indicates that the node should not!
  • Such a bug would be difficult to track down, since misleading status logs would indicate the node is not syncing, when it actually is

Solution

• Introduce logic that prevents a node from syncing if the platform status isn't in the permissible list

• Eventually, it would be good if we could always have gossip "started", and control whether or not we are syncing via platform status only

  • This would allow us to discard the racy gossip start/stop logic

[alittley]

Done