fix(gms): fence and supervise persistent KV services

[hutm]

Summary

Fence stale writer generations across restarts and supervise the persistent KV directory and helper services.

Why this is needed

Route removal and heartbeat loss detect failure but do not validate the metadata generation or guarantee that every recovery helper is healthy.

Validation

• The cumulative publication tip is byte-identical to the locally validated implementation.
• Cumulative focused validation: 994 passed, 1 skipped; the later vLLM-focused rerun passed 244 tests.
• Local TP2 crash tests passed exact-output recovery for vLLM and SGLang; final performance qualification remains tracked in DEP: Production hardening for persistent GMS KV failover #15035.
• This draft checkpoint still requires normal CI and reviewer validation before merge.

Stack

• Part 4 of 15 in the production-hardening train.
• Draft DEP and complete review order: DEP: Production hardening for persistent GMS KV failover #15035.
• Base: hutm/gms-hardening-03-tp-reattach.

This PR is intentionally draft and stacked on its immediate predecessor.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.