fix(gateway): log a cron store notice that fails to send

Gate finding C12 r3 #2 (quality): the notice listener dropped the future
from run_coroutine_threadsafe, so a send that raised (adapter error, a
profile scope that failed to bind) vanished without a trace. Keep the future
and log failures from a done-callback, the same way plugin message injection
does in gateway/run_plugin_injection.py.

gateway/cron_store_notices.py

@@ -9,6 +9,7 @@ send path the state.db warning uses.
 from __future__ import annotations
 
 import asyncio
+import concurrent.futures
 import contextlib
 import logging
 import threading
@@ -45,13 +46,23 @@ def install_cron_store_notices(runner, loop: asyncio.AbstractEventLoop) -> None:
             else:
                 return  # its outage was never announced: no lone "recovered"
             last_sent[store] = now
-        asyncio.run_coroutine_threadsafe(send_cron_store_notice(runner, event, record), loop)
+        future = asyncio.run_coroutine_threadsafe(send_cron_store_notice(runner, event, record), loop)
+        future.add_done_callback(lambda done: _log_notice_failure(done, event, store))
 
     set_transition_listener(on_transition)
     for record in degraded_records():
         on_transition("unwritable", record)
 
 
+def _log_notice_failure(done, event: str, store: str) -> None:
+    try:
+        done.result()
+    except (asyncio.CancelledError, concurrent.futures.CancelledError):
+        return
+    except Exception:  # the ticker thread already moved on: log, never raise
+        logger.warning("Cron store %s notice for %s failed", event, store, exc_info=True)
+
+
 def _profile_home(profile):
     from hermes_constants import get_routing_process_hermes_home
     if profile is None:

tests/cron/test_due_scan_save_failure.py

@@ -225,7 +225,7 @@ def test_unwritable_store_degrades_once_throttles_and_catches_up_once(cron_store
 
 @pytest.mark.platforms("posix")  # POSIX mode bits; root ignores them
 @pytest.mark.skipif(hasattr(os, "geteuid") and os.geteuid() == 0, reason="root bypasses directory modes")
-def test_unwritable_store_is_shown_in_cron_status_and_announced_once(cron_store, monkeypatch, capsys):
+def test_unwritable_store_is_shown_in_cron_status_and_announced_once(cron_store, monkeypatch, capsys, caplog):
     """`hermes cron status` probes the store itself (a real 0500 dir) and leads with the red
     headline + fix; the gateway posts ONE home-channel notice on entry and ONE on recovery,
     each naming the store, the error, since-when and the skipped runs; re-entry within the hour
@@ -267,9 +267,12 @@ def test_unwritable_store_is_shown_in_cron_status_and_announced_once(cron_store,
         save_jobs([_due_job()])
         cronjobs.warn_store_unwritable(enospc, "x", "scan", [_due_job()])  # flaps back within the hour
         loop.run_until_complete(asyncio.sleep(0.05))
+        install_cron_store_notices(SimpleNamespace(), loop)  # a send that raises is logged, not lost
+        loop.run_until_complete(asyncio.sleep(0.05))
     finally:
         loop.close()
     assert len(sent) == 2
     assert str(cron_dir) in sent[0] and "ENOSPC: No space left on device" in sent[0] and "since " in sent[0]
     assert f"fix permissions on {cron_dir}" in sent[0]
     assert "writable again; 1 skipped run(s), catching up once per job" in sent[1]
+    assert any(r.levelname == "WARNING" and "unwritable notice for" in r.getMessage() for r in caplog.records)