fix(vision-manager): retain analysis-enabled gate alongside memory-yield

The analysis-disabled gate (stop/inhibit vision when Admin→Settings analysis
is OFF, so stale queued jobs don't pin ~4.4GiB) was applied directly on the
host and never committed, so it was absent from main and from this branch —
merging as-is would have regressed it. Re-introduce it here so the merged
manager has BOTH gates: vision is held down if analysis is disabled OR host
memory is tight. Also grant the read-only role SELECT on app_settings.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

vision-manager/manager.sh

@@ -64,6 +64,20 @@ crawl_running() {
         AND state = 'running';" 2>/dev/null || echo "ERR"
 }
 
+# Is the analysis subsystem enabled in Admin → Settings? The runtime source of
+# truth is the `app_settings` row (key='analysis'), NOT the .env seed. When
+# analysis is OFF the worker drains nothing, so leftover pending jobs would
+# otherwise keep vision pinned forever — we must gate on this too, not just the
+# queue depth. Returns the literal 'true'/'false'; 'ERR' (or anything else) on
+# query failure is treated as enabled by the caller (fail-open: never yank
+# vision out from under a live analysis on a transient DB hiccup).
+analysis_enabled() {
+  "${PSQL[@]}" -c \
+    "SELECT CASE WHEN COALESCE((value->>'enabled')::boolean, false)
+                 THEN 'true' ELSE 'false' END
+       FROM app_settings WHERE key = 'analysis';" 2>/dev/null || echo "ERR"
+}
+
 # ---- Memory pressure -------------------------------------------------------
 # Host-wide used% from MemAvailable (NOT MemFree, which sits near "full" thanks
 # to page cache and would false-trigger constantly). /proc/meminfo in this
@@ -151,6 +165,7 @@ log "started: container=$VISION_CONTAINER health=$VISION_HEALTH_URL poll=${POLL_
 idle_for=0        # seconds the queue has been empty while vision is running
 up_for=0          # seconds vision has been running (for MAX_UPTIME backstop)
 crawl_deferred=0  # 1 while a start is held off by the crawl mutex (log on flip only)
+analysis_off=0    # 1 while analysis is disabled in settings (log on flip only)
 mem_yield_until=0 # epoch deadline of the active memory-yield cooldown (0 = none)
 mem_inhibited=0   # 1 while starts are inhibited by memory pressure (log on flip only)
 
@@ -169,6 +184,22 @@ while true; do
       ;;
   esac
 
+  # Gate on the runtime enable flag. If analysis is OFF, force the backlog to 0
+  # so the normal idle path runs the STOP_DEBOUNCE timer and shuts vision down.
+  # We deliberately reuse the idle debounce (rather than stopping instantly) so
+  # a quick toggle off→on inside the debounce window keeps vision warm — no
+  # cold reload for a momentary flip. Fail-open: only a literal 'false' counts
+  # as disabled; 'ERR'/empty leaves the real queue depth in effect.
+  if [ "$(analysis_enabled)" = "false" ]; then
+    if [ "$analysis_off" != "1" ]; then
+      log "analysis disabled in settings — ignoring ${pending} queued job(s); idle timer will stop vision"
+      analysis_off=1
+    fi
+    pending=0
+  else
+    analysis_off=0
+  fi
+
   # Memory-yield gates (before the backlog logic). The active stop handles a
   # running vision the idle path can't; the inhibit forces pending=0 so the
   # existing idle-debounce path keeps it down (same shape as the crawl mutex).

vision-manager/readonly-role.sql

@@ -26,7 +26,11 @@ SELECT 'CREATE ROLE vision_manager LOGIN'
 -- (Re)set the password every run so rotating it is just a re-apply.
 ALTER ROLE vision_manager LOGIN PASSWORD :'pw';
 
--- CONNECT to the current database, and read-only on the one table we poll.
+-- CONNECT to the current database, and read-only on the two tables we poll:
+-- crawler_jobs (backlog depth + crawl-in-flight) and app_settings (the runtime
+-- analysis-enabled flag — without this the manager can't tell that analysis was
+-- turned off and would keep vision pinned on stale queued jobs).
 GRANT CONNECT ON DATABASE :"DBNAME" TO vision_manager;
 GRANT USAGE ON SCHEMA public TO vision_manager;
 GRANT SELECT ON crawler_jobs TO vision_manager;
+GRANT SELECT ON app_settings TO vision_manager;