fix(night-shift): Remove poll timeout that aborted Explorer runs

The polling loop in agentic triage raised a TimeoutError after 120s,
which abandoned the entire night shift run for an org even though the
Explorer was still working on the Seer side. For large orgs like Sentry
the Explorer can take longer than 120s to investigate all candidates.

Remove the client-side poll timeout and let the task's own processing
deadline (5 minutes) be the backstop instead.

Fixes SENTRY-5NFV
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: trevor-e

src/sentry/tasks/seer/night_shift/agentic_triage.py

@@ -131,7 +131,6 @@ def _triage_candidates(
 
 
 POLL_INTERVAL = 2.0
-POLL_TIMEOUT = 120.0
 
 
 def _poll_with_logging(
@@ -192,10 +191,6 @@ def _poll_with_logging(
             )
             return state
 
-        elapsed = time.monotonic() - start_time
-        if elapsed >= POLL_TIMEOUT:
-            raise TimeoutError(f"Explorer run {run_id} timed out after {POLL_TIMEOUT}s")
-
         time.sleep(POLL_INTERVAL)