Propagate Spark time zone to DataFusion

[linhr]

Right now a few data sources (rate and binaryFile) use execution.time_zone in the DataFusion session config to determine the time zone for the timestamp data type in the output schema. But the Spark session time zone is not propagated to DataFusion session config. We should fix this. This affects the result of .show().