feat(podcasts): feed poller + podcast_poll_interval_seconds setting (task 1.6)

One ticker re-reading its interval from the settings registry each
cycle (live tunable — the cache stays honest because writes go through
the settings API path, AGENTS.md gotcha #2). Per tick: list feeds
(first-admin identity, invariant 12), skip is_complete feeds (back
off; explicit refresh still works), poll each (conditional GET via
stored validators), and apply the subscriber auto-download policy
union after each successful fetch so new episodes queue without
waiting a tick.

Schema/registry: podcast_poll_interval_seconds (default 3600s, floor
300s, category podcasts) — seeded with ON CONFLICT DO NOTHING per the
idempotency rule.

database/schema/schema.sql

@@ -100,7 +100,9 @@ INSERT INTO system_settings (setting_key, setting_value, description, setting_ty
 ('sync_queue_interval_seconds', '5', 'How often the sync queue flushes', 'int', '1', '3600', TRUE, 'sync'),
 ('sync_queue_batch_size', '50', 'Maximum items processed per sync queue flush', 'int', '1', '10000', TRUE, 'sync'),
 ('worker_pool_size', '3', 'Number of background worker goroutines', 'int', '1', '100', TRUE, 'performance'),
-('worker_queue_cap', '100', 'Background worker job queue capacity', 'int', '1', '10000', TRUE, 'performance')
+('worker_queue_cap', '100', 'Background worker job queue capacity', 'int', '1', '10000', TRUE, 'performance'),
+-- podcasts
+('podcast_poll_interval_seconds', '3600', 'How often to poll podcast feeds for new episodes (seconds; complete feeds back off)', 'int', '300', '604800', FALSE, 'podcasts')
 ON CONFLICT (setting_key) DO NOTHING;
 
 -- Create refresh_tokens table