Here's my solution

I've assumed that the number of threads is optimal, when the addition of a new one improves the execution time of the algorithm by less than 5%. you can change this value by modifying STOP_CONDITION variable.

I've also changed head_block = blockchain.get_current_block_num()-int(sys.argv[1]) to head_block = blockchain.get_current_block_num(), which I believe is correct