v1.52.0

Bug fixes surfaced by an internal audit pass, plus doc corrections.

Fixed

• Monitor stopped alerting for containers that recover by restarting. The v1.51.0 health debounce only ended an unhealthy episode on a direct unhealthy → healthy transition. A container that heals by restarting goes unhealthy → starting → healthy, which never matched — so no recovery alert fired and, worse, that container's future unhealthy episodes were silently suppressed for good. An episode now ends whenever a container leaves the unhealthy state, and stale state is pruned for vanished containers.
• False "update available" against non-conformant registries. A 200 manifest response with no Docker-Content-Digest header returned an empty digest that slipped past the reachability check and read as an update. Now treated as a failed lookup, matching the self-update path.
• DEBUG=true as an environment variable now works. It was documented and advised but never actually read from the environment — only the /debug command / Web UI toggle set it. It now seeds the initial debug state like every other setting.

Docs

• Corrected the command count and added the missing commands to the table; clarified self-update (Docksentry updating itself) vs auto-updating your other containers; narrowed the "everything persists via the Web UI" claim to what actually does; surfaced the /events command and Container Events history.