How do you store checkpoints and start from those offsets while reading data from eventhub using azure-eventhub-spark in python, without taking care of partitions explicitly? I