Add ability to control process redistribution during operation.

[veedo]

Problem

During Node-Up and Node-Down events there is a lot of activity.
For example, I have :syn running as my registry and as an event bus, and on node change there is a flurry of activity where it syncs the process groups and registry. Unfortunately this is also when Horde decides to redistribute processes.
The result is that some calls time out.

Solution

I am adding a feature to control the redistribution during operation and would like some opinions on which direction to go.

Options

Option 1:

Added a "force_redistribute" function which sends messages to all the nodes to trigger redistribution
Added an optional argument to:
defp handoff_processes(state, force_redistribution \\ false) do
Downside: May need to call it multiple times since it isn't using the crdt

Option 2:

Add Application.get_env call to "handoff_processes" that is checked in addition to :process_redistribution
Downside: The env variable is global and affects all supervisors

Option 3:

Add function/message to modify the :process_redistribution or add a new flag to the state that gets checked in addition to :process_redistribution
Downside: One more flag to keep track of

Option 4:

Add :manual to :process_redistribution flag
Add entry to the CRDT mirrors a state.redistribute_children flag
Trigger "handoff_processes" on a CRDT diff message for the flag
"handoff_processes" checks state.redistribute_children when set to :manual
Downside: ?