[checkpoint] Run syncing in active process for fullnodes

[williampsmith]

This adopts a similar pattern to the handling of the sui-node gossip process handle (which, for full nodes is a process to sync state from a quorum of validators, and for validators, is an active gossip process for txes, etc.), but for the checkpoint process.

Before:

• Full nodes do a full checkpoint sync once before spawning an active process for syncing state with validators
• Validator nodes spawn an active process for syncing and driving checkpoints

After:

• Both full nodes and validators spawn an active checkpoint process.
• For validators, this is the same checkpoint driving process as before
• For full nodes, this is an active checkpoint syncing process that runs periodically

Other notes on this change:

• In order to facilitate testing checkpoints on a simtest swarm, it was necessary to make CheckpointProcessControl configurable so as to avoid waiting 2+ minutes for the pause between checkpoint times. CheckpointProcessControl is now configured by setting a member of the Node, and for simtests that call start_a_fullnode*, is configured to wait on the order of seconds between checkpoints.
• As such, we can also now configure CheckpointProcessControl from the NetworkConfig yaml