feat(watchdog): sidecar-watchdog F1b — monitoring brain in a separate container (ORCH-100)

Add the `watchdog/` package (thin Python-3.12 stdlib-only daemon) and the
`orchestrator-watchdog` compose service — the brain half of the domain-0
observability pair. F1a (ORCH-099) exposes GET /metrics raw signal; F1b reads it,
augments with host / container / dependency probes, runs each signal through a
generalised pure decision function (decide(signal_active, prev, now, cooldown),
a strict superset of disk_watchdog.decide_action) with per-signal in-memory
dedup/throttle/recovery, and alerts over its OWN independent Telegram channel.

Key properties (ADR-001):
- Observer separated from observed: separate container; /metrics not answering is
  itself the master `orch_down` alarm (debounced K ticks — no flap on a hiccup).
- Strictly read-only: docker.sock GET-only + mounted :ro (double guard), host
  paths :ro, no DB/disk writes, no process control — self-hosting-safe.
- never-raise on three levels (per-source/per-tick/per-send) + WATCHDOG_ENABLED
  kill-switch (disabled -> inert idle-loop, not exit).
- Disk anti-duplicate (D6): disk_watchdog (ORCH-063) stays sole owner of the 85%
  alert; sidecar carries orch_down + an opt-in 97% ceiling (default off).
- NO import from src/** (C-1); src/**, STAGE_TRANSITIONS, QG_CHECKS, check_*, DB
  schema — untouched. env_file optional so a missing .env.watchdog never breaks
  `docker compose up` for the prod orchestrator.

Tests: tests/watchdog/ (TC-01…TC-13) + full tests/ regression green (TC-14).
Docs: CHANGELOG, .env.example canon (WATCHDOG_*); architecture README + adr-0033
authored at the architecture stage.

Refs: ORCH-100

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

.env.example

@@ -465,3 +465,48 @@ ORCH_POST_DEPLOY_BASE_URL=http://localhost:8500
 # DB title TEXT is unbounded). Default 200. An invalid/empty value gracefully
 # degrades to 200 (the process never crashes on startup).
 ORCH_QG0_TITLE_MAX=200
+
+# ── ORCH-100 (FND/F1b): sidecar-watchdog (orchestrator-watchdog container) ─────
+# The monitoring brain runs in a SEPARATE container with its OWN config. These
+# keys are read by the watchdog package (watchdog/config.py), NOT by the
+# orchestrator. At runtime they live in `.env.watchdog` (env_file of the
+# orchestrator-watchdog service); this block is the canon. NO real secrets here.
+#   ENABLED        -> kill-switch; false (or not starting the service) -> inert.
+#   INTERVAL_S     -> seconds between ticks.
+#   HTTP_TIMEOUT_S -> per-request timeout (metrics / pings / docker / telegram).
+#   COOLDOWN_S     -> re-alert throttle for a sustained signal (anti-spam).
+#   METRICS_URL    -> orchestrator /metrics (host-network -> 127.0.0.1:8500).
+#   ORCH_DOWN_TICKS-> K consecutive /metrics failures before "орк не отвечает".
+#   MEM_PCT        -> host memory used-% threshold.
+#   DISK_CRIT_*    -> OPT-IN independent disk CEILING (disk_watchdog/ORCH-063 owns
+#                     the 85% alert; this is a higher ceiling on the sidecar's own
+#                     channel, OFF by default -> no double disk-alert, AC-5/D6).
+#   DISK_PATHS     -> host paths measured for the opt-in ceiling.
+#   AGENT_HUNG_MIN -> runtime minutes before an agent with ~0 CPU is "hung".
+#   AGENT_CPU_FLOOR-> CPU fraction below which a long-running agent counts as hung.
+#   STAGE_STUCK_MIN-> minutes a task may sit in one stage before alerting.
+#   QUEUE_DEPTH    -> queued-job depth threshold.
+#   CONTAINERS     -> CSV of container names to watch (status != running/healthy).
+#   DOCKER_SOCK    -> path to the read-only docker.sock inside the container.
+#   DEPS           -> CSV of name=url dependency pings (empty -> no pings).
+#   TG_BOT_TOKEN / TG_CHAT_ID -> the sidecar's OWN Telegram bot/chat (independent
+#                     of the orchestrator's; absent -> logs, does not send).
+WATCHDOG_ENABLED=true
+WATCHDOG_INTERVAL_S=30
+WATCHDOG_HTTP_TIMEOUT_S=5
+WATCHDOG_COOLDOWN_S=1800
+WATCHDOG_METRICS_URL=http://127.0.0.1:8500/metrics
+WATCHDOG_ORCH_DOWN_TICKS=3
+WATCHDOG_MEM_PCT=90
+WATCHDOG_DISK_CRIT_ENABLED=false
+WATCHDOG_DISK_CRIT_PCT=97
+WATCHDOG_DISK_PATHS=/repos,/app/data
+WATCHDOG_AGENT_HUNG_MIN=20
+WATCHDOG_AGENT_CPU_FLOOR=0.01
+WATCHDOG_STAGE_STUCK_MIN=120
+WATCHDOG_QUEUE_DEPTH=20
+WATCHDOG_CONTAINERS=orchestrator
+WATCHDOG_DOCKER_SOCK=/var/run/docker.sock
+WATCHDOG_DEPS=
+WATCHDOG_TG_BOT_TOKEN=
+WATCHDOG_TG_CHAT_ID=