VIOLET fix: connect the divergence driver's data feed (dark + observe paths)

The driver owned a freshly built HazelcastDataFeed that nothing ever
connected; every 1s poll hit features_map=None ('NoneType' .get) at
ERROR level and no scans reached the divergence monitor. Connect with
retry before sampling. Verified live: WS bookTicker up for 50 symbols,
divergence rows landing in real time.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

prod/launch_dolphin_violet.py

@@ -223,6 +223,19 @@ def _build_divergence(sink=None):
 
 async def _divergence_driver(divergence, data_feed, poll_s: float) -> None:
     """Shared scan-sampling loop for both DARK and observe modes."""
+    # The driver owns this feed instance and is its only connector; an
+    # unconnected HazelcastDataFeed has features_map=None and every poll
+    # raises 'NoneType' has no attribute 'get' at ERROR level (1 Hz).
+    while True:
+        ok = False
+        try:
+            ok = await data_feed.connect()
+        except Exception as exc:  # noqa: BLE001 — sampling must never die
+            LOGGER.warning("divergence data feed connect error: %s", exc)
+        if ok:
+            break
+        LOGGER.warning("divergence data feed not connected; retrying in 5s")
+        await asyncio.sleep(5.0)
     started = False
     symbol = os.environ.get("DOLPHIN_VIOLET_SNAPSHOT_SYMBOL", "BTCUSDT")
     mode = os.environ.get("DOLPHIN_VIOLET_VENUE_MID_MODE", "ws").lower()