Add OPERATIONAL SAFEGUARDS headers, skip prefill during active rescans

arr_full_rescan.sh, arr_cache_prefill.sh, and arr_rescan_monitor.sh were
missing the standard SAFEGUARDS header section other Arrs_Stack/Tools
scripts have. Also: arr_cache_prefill.sh now checks for an active rescan
before fetching, instead of doing a live fetch that arr_cache_write()
would just refuse to persist anyway -- avoids wasted API calls every
30min during a long rescan. arr_rescan_monitor.sh was also missing an
actual root check despite writing cache files; added it to match
convention rather than just document a safeguard that wasn't there.

Arrs_Stack/arr_cache_prefill.sh

@@ -6,11 +6,18 @@
 # PURPOSE
 # ─────────────────────────────────────────────────────────────────────────────
 # Populates the shared tracked-data cache (see arr_get_tracked_data() in common.sh) for
-# Lidarr, Sonarr, and Radarr once at array start, before any other script needs it. Without
-# this, each arr's cache stays cold from boot until whichever script happens to touch that
-# arr first writes through — which could be hours, depending on the daily schedule. One-shot:
-# runs and exits. Originally Lidarr-only (lidarr_cache_prefill.sh, 2026-07-16), generalized
-# the same day to cover all three arrs once the cache mechanism itself was generalized.
+# Lidarr, Sonarr, and Radarr. Without this, each arr's cache stays cold until whichever
+# script happens to touch that arr first writes through — which could be hours, depending
+# on the daily schedule. Originally Lidarr-only (lidarr_cache_prefill.sh, 2026-07-16),
+# generalized the same day to cover all three arrs once the cache mechanism itself was
+# generalized.
+#
+# Runs on two schedules (2026-07-17): once at array start (closes the cold-boot gap, 10min
+# wait ceiling for slow-starting containers) and again every 30min via
+# CRITICAL_MAINTENANCE_SCRIPTS with a 1min wait ceiling (ARR_PREFILL_WAIT_MINUTES=1 override)
+# — a live fetch+write takes seconds, so there's no reason to tolerate the boot-time wait on
+# a recurring job. This is what keeps the cache-first consumers' data reliably under 30min
+# old instead of only refreshing whenever some other script happens to write through.
 #
 # ==============================================================================================
 # OPERATIONAL MODEL
@@ -23,6 +30,22 @@
 # An arr not configured on this host (e.g. Lidarr is HOST1-only) is skipped cleanly.
 #
 # ==============================================================================================
+# OPERATIONAL SAFEGUARDS
+# ==============================================================================================
+#
+#   Root required       — chown-free here, but matches convention across Arrs_Stack/
+#   jq required         — skips the whole run cleanly (not fatal) if jq is missing
+#   acquire_lock        — prevents two invocations of this script overlapping (default strict
+#                         mode) — matters if the boot-time run is still waiting on a slow
+#                         container when the first 30min critical-tier tick fires
+#   Reachability retry  — tolerates a slow-starting container up to ARR_PREFILL_WAIT_MINUTES;
+#                         never fatal if one never comes up, that arr's cache just stays cold
+#   Active-rescan check — skips an arr this cycle if a rescan-type command is running, instead
+#                         of a live fetch arr_cache_write() would refuse to persist anyway
+#                         (2026-07-17) — avoids wasted API calls during a long rescan
+#   Per-arr isolation   — one arr failing or timing out never blocks or fails the others
+#
+# ==============================================================================================
 # CONFIGURATION
 # ==============================================================================================
 #
@@ -33,7 +56,10 @@
 #   All aliased by detect_hosts()
 #
 # master.conf
-#   ARR_PREFILL_WAIT_MINUTES — how long to retry reaching each arr before giving up on it (default 10)
+#   ARR_PREFILL_WAIT_MINUTES — how long to retry reaching each arr before giving up on it
+#   (default 10). The CRITICAL_MAINTENANCE_SCRIPTS entry overrides this to 1 via parse_args'
+#   VAR=VAL mechanism for the 30min recurring run — the 10min default is sized for cold boot,
+#   not a job that fires every half hour.
 #
 # ==============================================================================================
 
@@ -78,6 +104,17 @@ _prefill_one() {
         (( waited += 15 ))
     done
 
+    # Skip cleanly if a rescan-type command is already active — arr_cache_write() below
+    # would refuse to persist a mid-rescan snapshot anyway (2026-07-17 guard), so fetching
+    # it live first would just be a wasted API call every time this fires during a long
+    # rescan. Matches the same check arr_full_rescan.sh and the cleanup scripts already do.
+    local active_cmd
+    active_cmd=$(arr_active_rescan_command "$arr_type" "$url" "$api_key" "$api_version")
+    if [[ -n "$active_cmd" ]]; then
+        info "${arr_type^} mid-rescan ($active_cmd) — skipping this cycle, cache stays as-is"
+        return 0
+    fi
+
     local endpoint="${ARR_LIBRARY_ENDPOINT[$arr_type]:-}"
     if [[ -z "$endpoint" ]]; then
         warn "No library endpoint known for ${arr_type} — skipping"