Trickle in validators

[sdbondi]

Problem

A large enough number of validators register for an epoch to cause the new validators entering in to exceed the previous BFT majority could cause a disruption in conseusus.

Proposal 1

"Trickle in" previously registered but inactive validators at the start of each epoch.

A validator registers and is added to the back of a queue. At the start of a new epoch, $n$ validators from the front become part of the active set.

Acceptance:

1. Register e.g. 50 validators, only $n$ become active at each epoch until all 50 are in the active set
2. $n$ is configurable
3. Tari swarm should still be able to start a new swarm with any amount of validators - after that, any new validators would be queued as normal. To implement this you could specify a minimum amount of validators before the queuing starts e.g. MIN_VALIDATOR_SET_QUEUING_SIZE=50.