fix(scheduler): trumpstruth poller never fires — APScheduler bug

next_run_time=None means "paused", not "default". The fallback poller was
registered but never triggered, leaving the system on CNN archive only.
Set explicit start time = now + half the poll interval so the two pollers
offset and don't hit upstream simultaneously.

Verified via /api/health/deep: trumpstruth.last_poll was null on the live
server before this fix despite the poller having been deployed.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

app/main.py

@@ -69,6 +69,11 @@ async def lifespan(app: FastAPI):
     # same external_id hash, so dedup is automatic. Whoever sees the post
     # first wins. Offset by half the interval so the two pollers don't
     # hammer the network at the same instant.
+    # Offset the fallback by half the interval so the two pollers don't hit
+    # upstream at the same instant. APScheduler treats next_run_time=None as
+    # "paused" (job never fires) — must be an actual datetime.
+    from datetime import datetime as _dt, timezone as _tz, timedelta as _td
+    offset = settings.truth_social_poll_seconds // 2
     _scheduler.add_job(
         poll_trumpstruth,
         "interval",
@@ -77,7 +82,7 @@ async def lifespan(app: FastAPI):
         id="trumpstruth_poll",
         max_instances=1,
         coalesce=True,
-        next_run_time=None,  # APScheduler will pick a fresh slot
+        next_run_time=_dt.now(_tz.utc) + _td(seconds=offset),
     )
     _scheduler.start()
     logger.info(