No, the scale limit is not increased from a single-node cluster to a three-node cluster. The scale limit remains the same unless you are using three DN2-APL-HW-XL appliances.
As mentioned, all three nodes are operating at the same time. The only difference is HA allows us to distribute the workload of the services among the three nodes and replicate the information for stateful services/databases, such as postgres, across the cluster. Each stateful service will have a primary instance working on a single node and secondary/standby instances working on the other nodes. But the primary instances for each stateful service may not all reside on the same node. For instance, postgres-0 may be the primary and postgres-1/2 will be the secondary instances. Postgres-0 may be on node 1, meaning node 1 is running the primary instance. Then we have Mongodb, which has three instances. Mongodb-1 may be the primary and Mongodb-0/2 will be the secondary. Mongodb-1 may reside on node 2, making that the primary instance holder.
I state all of this to show an example of how the nodes work together to keep the cluster functioning. In the event that one node fails, the stateless services should redistribute to another working node while the stateful services should go into a PENDING or NODE_LOST state. Then if that node that went down held a primary instance, one of the secondary instances should promote to primary and take over write privileges.
