fix(sidecar_init): log after spawn in start_all to close SIGTERM race

`start_all()` was logging "starting {name}" before appending the
process to `self.procs`, so a SIGTERM arriving between the log write
and the append would call `request_shutdown()` with the new daemon
absent from `self.procs` and therefore never forwarded SIGTERM.
Moving the log after the append eliminates the window.

Signal handlers are also moved before `sup.start_all()` in `main()` so
they are registered before any child is spawned; previously there was a
window where SIGTERM used the default handler (silent process death) if
it arrived between `start_all()` returning and `signal.signal()`.

`TestMainEndToEnd._run` is updated to use a reader thread and a
`wait_for_output` synchronisation barrier instead of a fixed sleep, so
`test_sigterm_clean_shutdown` does not race against slow CI startup time.

bot_bottle/sidecar_init.py

@@ -167,8 +167,8 @@ class _Supervisor:
 
     def start_all(self) -> None:
         for spec in self.specs:
-            _log(f"starting {spec.name}")
             self.procs.append((spec, _spawn(spec)))
+            _log(f"starting {spec.name}")
 
     def request_shutdown(self, reason: str) -> None:
         if self.shutdown_at is not None:
@@ -353,8 +353,6 @@ def main(argv: Sequence[str] | None = None) -> int:
         return 0
 
     sup = _Supervisor(specs)
-    sup.start_all()
-
     signal.signal(signal.SIGTERM, lambda *_: sup.request_shutdown("SIGTERM"))  # type: ignore
     signal.signal(signal.SIGINT, lambda *_: sup.request_shutdown("SIGINT"))  # type: ignore
     # SIGHUP reload path: egress_apply.py runs `docker kill
@@ -362,6 +360,7 @@ def main(argv: Sequence[str] | None = None) -> int:
     # delivers SIGHUP to PID 1 (this supervisor); forward it to
     # mitmdump so it reloads its addon.
     signal.signal(signal.SIGHUP, lambda *_: sup.forward_signal(signal.SIGHUP, "egress"))  # type: ignore
+    sup.start_all()
 
     while not sup.tick():
         time.sleep(_POLL_INTERVAL)