v2.5.1
Bug fix
Fetcher thread killed by its own safety monitor
The passive safety system's cooperative restart mechanism was terminating the fetcher thread on every cycle:
enable_graceful_degradation_if_needed()reducedg_pb_interval_secondsfrom 3600 to 1800 (0 active calls)- This shrunk the hang-detection timeout below the fetcher's sleep duration
- The fetcher was flagged as hung and exited permanently — never restarted
Fix: Removed the fetcher from passive safety monitoring entirely. A thread that sleeps for an hour by design is not a candidate for hang detection. Net -53 lines.