slow array/metadata/sub_group creation with GCS

[AndrewAnnex]

I am using zarr 2.18.3 on google cloud to create empty zarr groups with seeded metadata and arrays in the following manner:

group = zarr_archive.create_group('foo')
for thing, array in list_of_stuff:
     sub_group = group.require_group(unique_name_per_thing)
     sub_group.attrs['bar'] = 'foo'
     gcarr = sub_group.zeros('baz', shape=(10000,100))
     gcarr = array
     smaller_group = sub_group.require_group('faz')
     # add a few atts and arrays to this new sub group

Unfortunately the performance isn't very good, at about 5 seconds per iteration.

Without simply writing to the local disk and manually syncing to google cloud storage, is there a more performant method to be attempted here? I think I've seen references to async methods but I haven't seen a worked out example in python in the docs, or I could use the parallel methodology but I am not sure if it's possible to pickle the top level group object, but maybe I could get that top level group made and then re-open the group with zarr.open?

tips and ideas would be appreciated