Allow Collection's update() and map() method to take dask arguments

[robin-cls]

When mapping a callback to the zcollection partitions, we might want to specify the 'resources' parameter when submitting the tasks.

zcollection.Collection.update:

    batches: Iterator[Sequence[str]] = dask_utils.split_sequence(
        selected_partitions, npartitions
        or dask_utils.dask_workers(client, cores_only=True))
    storage.execute_transaction(
        client, collection.synchronizer,
        # client.map(local_func, batches, resources=resources, key=func.__name__)) CURRENT CODE
        client.map(local_func, batches, resources=resources, key=func.__name__)) # REQUESTED CHANGE

In case we parallelize a long running task, dask worker may declare itself as idle if the blocking task releases the GIL or is blocked doing IO. Setting resources can help us limit overscheduling due to this behavior.