Add full banner headers to all scripts across the codebase

Every script now has the established header format: PURPOSE with ─────── separator,
OPERATIONAL MODEL, DESIGN PRINCIPLES, OPERATIONAL SAFEGUARDS, CONFIGURATION, and
RUNTIME MODES — structured with full ====== banner sections throughout.

Orchestrators converted from compact ── inline format to full banners. Stale
emby-fallback and dirty sync references removed from Plugin/user_script_plug-in.sh.

Monitors/emby_session_report.sh

@@ -18,6 +18,25 @@
 # configuration issue. Silent on clean runs.
 #
 # ==============================================================================================
+# DESIGN PRINCIPLES
+# ==============================================================================================
+#
+# No Persistent State
+#   Every report is generated fresh from the Emby API. No local database, no
+#   incremental tracking. A missed run leaves no gap — the next run simply
+#   covers its own window.
+#
+# Silent When Healthy
+#   The report goes to Discord/notification as a summary. Transcode alerts are
+#   the only proactive notification — high transcode ratios may indicate a
+#   misconfigured client that needs attention before it becomes a performance issue.
+#
+# Section Independence
+#   Each report section (sessions, library, activity, top content) guards its own
+#   API calls. A failure in one section does not abort the others — the report
+#   produces partial output rather than nothing.
+#
+# ==============================================================================================
 # OPERATIONAL SAFEGUARDS
 # ==============================================================================================
 #

Monitors/mesh_monitor.sh

@@ -24,6 +24,36 @@
 # Safe to share with mesh members — contains no API keys, passwords, or SSH keys.
 #
 # ==============================================================================================
+# DESIGN PRINCIPLES
+# ==============================================================================================
+#
+# Read-Only, No Network Calls
+#   All data comes from conf files — no SSH, no API calls, no pings. The output
+#   is always instant and never fails due to a node being unreachable. This makes
+#   it safe to run at any time without side effects.
+#
+# Scales Automatically
+#   Iterates all defined HOST* vars rather than a hardcoded list. Adding a new
+#   node to master.conf/host*.conf makes it appear in the output immediately.
+#
+# Safe to Share
+#   Output contains only identity and coverage configuration — no API keys,
+#   no passwords, no SSH keys. The report can be shared with other mesh members
+#   without exposing secrets.
+#
+# ==============================================================================================
+# OPERATIONAL SAFEGUARDS
+# ==============================================================================================
+#
+# No External Dependencies
+#   Reads only from already-sourced conf files. No curl, no ssh, no docker —
+#   nothing that can fail, hang, or require credentials.
+#
+# Empty Mesh Guard
+#   collect_hosts() populates ALL_HOST_IDS — if no HOST* vars are defined the
+#   output sections iterate over an empty array and exit cleanly.
+#
+# ==============================================================================================
 # CONFIGURATION
 # ==============================================================================================
 #

Monitors/smart_health.sh

@@ -19,6 +19,25 @@
 # from master.conf if dynamix.cfg is not found.
 #
 # ==============================================================================================
+# DESIGN PRINCIPLES
+# ==============================================================================================
+#
+# Live Queries, No Persistent State
+#   Every run queries smartctl directly — no cached attribute history, no trend
+#   tracking. Each report is a snapshot of current drive health. This keeps the
+#   script simple and the output always current.
+#
+# Threshold Parity With unRAID Dashboard
+#   Temperature thresholds are read from dynamix.cfg — the same values unRAID
+#   uses on its own dashboard. A consistent threshold means no conflicting alerts
+#   between this script and the built-in unRAID warnings.
+#
+# Silent When Healthy
+#   No output, no notification on a clean run. The absence of a report is the
+#   confirmation that all drives passed. Noise from weekly healthy runs would
+#   erode attention to the reports that matter.
+#
+# ==============================================================================================
 # OPERATIONAL SAFEGUARDS
 # ==============================================================================================
 #

Monitors/system_tuning_monitor.sh

@@ -18,6 +18,24 @@
 # WebGUI slowdowns or timeouts under load.
 #
 # ==============================================================================================
+# DESIGN PRINCIPLES
+# ==============================================================================================
+#
+# Snapshot for Trend, Not Just Alert
+#   Exhaustion events are rarely instant — they build over hours or days.
+#   Logging every 6 hours builds a trend that weekly_health_digest.sh can
+#   surface as a warning count, catching gradual pressure before it becomes
+#   an outage.
+#
+# Bounded Log Size
+#   Log entries are trimmed to TUNING_LOG_RETENTION days on every write.
+#   The log never grows unbounded regardless of how long the server runs.
+#
+# Silent When Healthy
+#   No output, no notification on a clean run. Threshold breach is the only
+#   signal — routine snapshots below the threshold produce nothing.
+#
+# ==============================================================================================
 # OPERATIONAL MODEL
 # ==============================================================================================
 #

Monitors/weekly_health_digest.sh

@@ -14,6 +14,24 @@
 # into a single digest. Reads only — writes nothing, changes nothing.
 #
 # ==============================================================================================
+# DESIGN PRINCIPLES
+# ==============================================================================================
+#
+# Aggregator, Not Generator
+#   This script reads state files that other scripts maintain. It never produces
+#   health data itself — it only presents what is already there. Each source
+#   script remains responsible for its own state; this script is the envelope.
+#
+# Profile-Driven Notification
+#   The cron schedule never changes. The DIGEST_PROFILE in master.conf controls
+#   when notifications actually send — switching from daily noise to weekly
+#   summaries is a one-line conf change, not a cron edit.
+#
+# Read-Only, No Side Effects
+#   Writes nothing, changes nothing, triggers nothing. Safe to run at any time
+#   for a health snapshot without affecting any running service or state file.
+#
+# ==============================================================================================
 # OPERATIONAL MODEL
 # ==============================================================================================
 #

Monitors/zfs_memory_snapshot.sh

@@ -15,6 +15,25 @@
 # comparison. In --dry-run mode, console only — nothing written to the log.
 #
 # ==============================================================================================
+# DESIGN PRINCIPLES
+# ==============================================================================================
+#
+# Informational, Not Interventional
+#   This script reports — it does not act. system_watchdog.sh handles
+#   threshold-based intervention. Keeping the roles separate means the report
+#   is never suppressed by the same logic that triggers remediation.
+#
+# Week-Over-Week Comparison
+#   Output is written to ZFS_REPORT_LOG so the same snapshot can be reviewed
+#   across weeks. Memory pressure and ARC creep are slow — a single run is
+#   rarely conclusive; the trend across weeks is what matters.
+#
+# Section Independence
+#   Each of the five report sections guards its own data source. ZFS not
+#   available, Docker not responding — those sections skip, the rest still run.
+#   A partial report is more useful than no report.
+#
+# ==============================================================================================
 # OPERATIONAL MODEL
 # ==============================================================================================
 #