fix: prevent email loop by detecting bounces and capping conversation length

- email_channel.py: add detect_automated_message() that inspects RFC 3834
  Auto-Submitted, mailer-daemon/postmaster sender patterns, X-Loop,
  multipart/report Content-Type, Precedence, and subject heuristics.
  fetch_unread_messages() now includes is_automated / automated_reason
  in every message dict.

- main.py: if is_automated is set, call agent.handle_automated_message()
  instead of process_message() — no reply is ever sent to a bounce source.

- agent/core.py: add MAX_USER_MESSAGES = 20 cap; process_message() returns
  "" without replying once a conversation exceeds the limit and calls
  notifier.notify_loop_escalation() on first breach.  New public method
  handle_automated_message() records the event and triggers the same
  one-shot admin alert.

- models/conversation.py: add loop_escalated: bool field (persisted) so
  the admin alert fires at most once per conversation.

- notifications/notifier.py: add notify_loop_escalation() which sends a
  plain-text warning to the admin CC list (Markus Graf / spielgruppen@).

https://claude.ai/code/session_01KwvR5hDPjSuJg4kvw5b5e5

main.py

@@ -98,6 +98,27 @@ def run_poll_loop(agent: EmailAgent, channel: EmailChannel, poll_interval: int)
             for msg in messages:
                 logger.info("Processing message from %s", msg["from"])
                 try:
+                    # ----------------------------------------------------------
+                    # Bounce / automated-sender guard
+                    # If the channel layer flagged this as an automated message
+                    # (bounce, out-of-office, delivery failure, …) we must NOT
+                    # reply — that would create or worsen an email loop.
+                    # Instead, alert the admin once and drop the message.
+                    # ----------------------------------------------------------
+                    if msg.get("is_automated"):
+                        logger.warning(
+                            "Automated/bounce message from %s — reason: %s — not replying",
+                            msg["from"],
+                            msg.get("automated_reason", "unknown"),
+                        )
+                        agent.handle_automated_message(
+                            sender_email=msg["from"],
+                            subject=msg["subject"],
+                            reason=msg.get("automated_reason", "automated sender detected"),
+                            inbound_message_id=msg["message_id"],
+                        )
+                        continue
+
                     # Prepend email headers so the LLM can extract the
                     # sender's address and subject (e.g. to fill in
                     # parentGuardian.email automatically).