Welcome to the Community,
if you ware able to power on the VMs in the HA cluster with Admission Control being configured to tolerate a host failure, then HA should be able to restart the VMs on the surviving hosts.
so basically the key here is to configure Admission Control properly so it can handle the fact that we will be over-provisioning CPUs at a rate of 3:1 ?
Basically yes, Admission Control will take the configured failures into account. However, performance may be degraded on the remaining hosts, depending on the remaining resources, and the workloads.