feat: add scalable kubeadm runtime worker pool

[convee]

Summary

• replace the fixed local topology with one kubeadm control plane and a 0..N gVisor Runtime worker pool
• add safe scale up/down, capacity synchronization, shared Lima networking, and stale-node cleanup
• move local Workspace persistence to CephFS RWX with pinned Rook and Ceph-CSI components
• update newcomer docs and the GitHub Pages homepage for the scalable standalone architecture

Validation

• make verify: 844 tests, 2 skipped; manifests, Console, and wheel passed
• make quickstart: passed in 59s on the isolated profile
• make e2e-local: 5/5 live suites passed in 60s
• live scaling: 1 to 2 workers, active-Runtime scale-down refusal, 2 to 1 to 0 to 1, retained worker disks, capacity 8/4/0/4
• two concurrent Runtime Pods scheduled across distinct workers and reported the gVisor kernel
• all sandbox Lima VMs stopped after validation