Fix coordinator memory leak

We were using a class attribute for saving API response data,
which causes corruption when there are more than one config
entries set up.

Move that to an instance attribute to avoid the collision.

This fixes #4

custom_components/teslafi/coordinator.py

@@ -19,7 +19,7 @@
 class TeslaFiCoordinator(DataUpdateCoordinator[TeslaFiVehicle]):
     """TeslaFi Update Coordinator"""
 
-    _vehicle = TeslaFiVehicle({})
+    _vehicle: TeslaFiVehicle
     data: TeslaFiVehicle
 
     _override_next_refresh: timedelta = None
@@ -30,6 +30,8 @@ def __init__(
         client: TeslaFiClient,
     ) -> None:
         self._client = client
+        self.data = None
+        self._vehicle = TeslaFiVehicle({})
         # TODO: implement custom Debouncer to ensure no more than 2x per min,
         #  as per API rate limit?
         super().__init__(
@@ -74,7 +76,7 @@ def _schedule_refresh(self) -> None:
     async def _refresh(self) -> TeslaFiVehicle:
         """Refresh"""
         current = await self._client.last_good()
-        LOGGER.info("Last good: %s", current)
+        LOGGER.debug("Last good: %s", current)
         self._vehicle.update_non_empty(current)
         last_remote_update = self._vehicle.get("Date")
         LOGGER.debug("Remote data last updated %s", last_remote_update)