fix: close and disconnect zmc's monitors when stopped during startup

[nabbi]

Automated PR, tested in a zm-holodeck sandbox. An AI agent (Claude Opus 5.5) reproduced
the bug on a disposable ZoneMinder built from upstream master c3d2e542b with
simulated cameras, wrote this fix and ran the checks below on the same lab. Minimal
human review before opening: please review it as you would any contributor's PR.

TL;DR

• Fixes: zmc stopped while it is still starting leaves its monitor threads running past logger and database teardown #5220: #: a SIGTERM during zmc startup skips Close()/disconnect() and leaves the monitor threads running past logger and database teardown.
• Change: drop the if (zm_terminate) break; after the startup loop; with zm_terminate set the capture loop doesn't run, so the existing Close()/disconnect() loop runs and the outer loop ends as on a normal stop; 1 file, +3/-1
• Risk: Close()/disconnect() now also run on monitors that never connected in this pass; both already handle that (reconnects run them on every monitor); stop times unchanged
• Tested: ThreadSanitizer, 20 stops during startup: 56 thread-leak reports and 13 logger exit races each before, none after; checks covering the diff green

Details: reproduction, change, checks

Before

Upstream master c3d2e542b, ThreadSanitizer build, 640x360 cameras; zmc -m 4 (ONVIF monitor)
started and stopped 0.5-3 s later, 20 times; reports from those processes:

  1     56  thread leak ... in std::thread::_M_start_thread(...)  [zmc]
  2     13  data race ../../../../src/libsanitizer/tsan/tsan_new_delete.cpp:150 in operator delete(void*, unsigned long)  [zmc]
  3     13  data race /src/src/zm_logger.cpp:350 in Logger::databaseLevel(int)  [zmc]
  4     13  data race /src/src/zm_logger.cpp:589 in logTerm()  [zmc]
  5     13  data race /src/src/zm_logger.cpp:80 in Logger::Logger()  [zmc]

Change

• src/zmc.cpp: remove the if (zm_terminate) break; after the startup loop (:328), with a
  comment saying why. The breaks inside the startup loop (:276, :312) only leave the for
  and now reach the Close()/disconnect() loop too. For a monitor that never got that far,
  Close() → Pause() finds no decoder, analysis thread, camera or event, and disconnect()
  returns early on mem_ptr == nullptr after clearing the zones (the same calls a reconnect
  makes on every monitor).
• Test: no unit test: the path is zmc's main(). Covered by the TSan probe above and holodeck's
  e2e/signals.py startup-stop scenario.

Behaviour changes

Change	Who sees it	Intended / collateral	Alternatives looked into	Chosen, why	Tested by
A zmc stopped during startup closes and frees its monitors before exiting	zmdc, logs	intended	Zone holding a non-owning pointer: frees the Monitor via ~Monitor → Close(), but a larger change touching every Zone user, and the loop is skipped anyway; Close()/disconnect() in a second loop after the outer loop: runs them twice on a normal stop	reach the existing loop	TSan probe; signals startup-stop PASS
A SIGHUP that arrived together with the SIGTERM during startup now also runs the reload block before exiting	nobody (exit follows)	collateral, same as a stop during capture today	guard the reload with !zm_terminate: a further change for no visible effect	unchanged
Stop time during startup	zmdc	no visible change			startup-stop delay/stop times below

Side fixes in this PR: none.

After

Branch fix/b65-zmc-stop-during-startup at 856cb250e (base c3d2e542b), tested in a merge with
two other fixes from the same lab session (combo = master + this +
fix/b55-usr-level-signal-safe + fix/b77-monitorlink-token-leak; no shared files):

"head": "69fa500a6683a16c937ae8be4d249a7eb52ba861", "branch": "combo", "dirty": false (ThreadSanitizer)
PASS  startup-stop: zmc -m 4 (onvif-1) started and stopped 20x (delay/stop time 0.8s/7.2s 2.6s/9.9s 2.4s/0.9s 1.1s/1.2s 1.7s/0.6s 1.6s/0.6s 2.1s/1.2s 2.5s/0.9s 0.7s/0.6s 0.6s/0.6s 2.6s/0.6s 1.6s/0.6s 2.4s/0.9s 0.5s/0.6s 1.6s/0.6s 2.3s/1.2s 1.1s/0.6s 2.9s/10.2s 2.8s/9.9s 0.6s/0.6s) [116s]
reports from those processes: no thread leak, no race on logTerm/Logger; only the known capture/decode races
(zm_monitor.cpp:3096, :3687, :2366, zm_packet.cpp:397/404, zm_monitor.h:1223), present on master too

Check	Result
holodeck checks for the diff (check-changed.py: unit-tests, script-tests, smoke, resilience, events, dblog --quick, signals --quick; plain build of combo)	unit-tests ok (14725 assertions in 251 test cases), script-tests PASS=31, smoke PASS=11 XFAIL=1 (stock zmonvif-probe.pl discovery, known), resilience PASS=5, events PASS=10, dblog --quick PASS=1 XFAIL=2 (known on master, #5197/#5208), signals --quick PASS=4 (hup-default, usr, usr-threads, startup-stop)
pyzmNg / ha-zoneminder / zmNinjaNg	not run: no API, web or stream change
contract baselines (master)	not run: no PHP change

Not run: the deep set. Backport to release-1.38: same code, the commit cherry-picks cleanly (not built).

Tested as 8cf52ca5c; 856cb250e is the same tree with fixes #5220 added to the message.

fixes #5220

🤖 Generated with Claude Code