The platform now survives losing a whole machine

The three machines now act as one cluster. If one is lost, the other two keep the platform running and hold the decision-making majority needed to do so safely.

Previously the platform depended on individual machines.