[Bug]: Cached thread rows keep claiming Working while their environment reconnects#25449
When an environment stops answering during a run, the client can show a reconnecting banner while its cached sidebar row still says Working with an increasing elapsed timer. This is observable even when Working shelf is enabled; a pinned thread remains in the pinned list by design.
Reproduction:
- Start a remote thread and optionally pin it.
- Interrupt the connection while its last received runtime is running.
- Wait for the environment connection presentation to become reconnecting.
- The sidebar continues to claim Working using cached runtime data.
Expected: distinguish unavailable connection from live work, suppress the live-work timer while disconnected, and restore the current run status after reconnect. Preserve run outcomes and pin/shelf placement; lost connectivity alone is not evidence of failure or completion. No elapsed-time cutoff is needed.
Scope: web/desktop sidebar and mobile thread-list status, using existing shared connection state. Investigating the server outage, recovery actions for orphaned threads, lineage redesign, and the separate mobile completion/shelf bug are outside this fix.
Confirmed on main 3ca62a5644572aab1cf7a058944aa44df516734c with deterministic failing status-resolver tests for web and mobile. Prior related reports: #4852 (closed), #4589 (subscription loss while transport stays connected); closed prior attempts #8194 and #8935. #24901 concerns authoritative completion updates and is distinct.