fix(mnemosyne): correct celery healthcheck command

The exec-form CMD has no shell, so $$HOSTNAME was never expanded,
causing "celery@$HOSTNAME" to match no node and every healthcheck to
fail. Remove the -d filter (one worker per container makes it
unnecessary) and add -t 8 to accommodate broker round-trip latency.

docker-compose.yaml

@@ -371,7 +371,12 @@ services:
     volumes:
       - media:/mnt/media
     healthcheck:
-      test: ["CMD", "celery", "-A", "mnemosyne", "inspect", "ping", "-d", "celery@$$HOSTNAME"]
+      # No -d destination: exec-form CMD has no shell, so $$HOSTNAME never
+      # expanded and the literal "celery@$HOSTNAME" matched no node → every
+      # check failed. There's one worker per container, so an unfiltered ping
+      # (any node replies = healthy) is correct. -t gives the reply room to
+      # round-trip through the broker on Oberon (~450ms observed) under jitter.
+      test: ["CMD", "celery", "-A", "mnemosyne", "inspect", "ping", "-t", "8"]
       interval: 60s
       timeout: 10s
       retries: 3