test: cover the node status watcher under synctest

[CMGS]

Test debt from the 2026-09 audit: the node status watcher had one test. These run it under testing/synctest, so the election retry, the keepalive expiry and the reconnect timer are exercised on the fake clock:

• withActiveLock cancels the work when the ephemeral lock expires and unregisters; a lost election is retried every keepalive/4 until it wins.
• run reconnects after the connection timeout and stops when the context ends.
• replayDeadJournals keeps ticking past a service status error and never takes over on one.
• initNodeStatus handles only unavailable nodes, at most nodeStatusHandlers at a time, and stops on a store error.
• monitor returns ErrMessageChanClosed when the stream closes and the context error when it ends.
• dealNodeStatusMessage sets workloads down only for dead nodes: alive and broken messages are dropped, a failed SetNode is swallowed.
• RunNodeStatusWatcher returns when the context ends.

Tests only, no production change. Gates: build, vet, full tests, lint, fmt-check, asl on linux and darwin green.