Benchmark configurations for building hnsw index

Currently, for full sync, we have a message:

NOTICE:  Hivesense 8 is attempting to process a block range: <95290001, 95300000>
NOTICE:  hnsw graph no longer fits into maintenance_work_mem after 1142920 tuples
DETAIL:  Building will take significantly more time.
HINT:  Increase maintenance_work_mem to speed up builds

It would be good to temporarily (for building the index) change work_mem