Fix worker count to use CPU affinity correctly and be properly bounded

We were using the `cpu_count`, which doesn't respect container affinity. Now, we use `cpu_affinity` and also bound to make sure we don't start a million workers

Fixes https://jira.coreos.com/browse/QUAY-928

conf/gunicorn_verbs.py

@@ -2,16 +2,16 @@ import sys
 import os
 sys.path.append(os.path.join(os.path.dirname(__file__), "../"))
 
-import multiprocessing
 import logging
 
 from Crypto import Random
 from util.log import logfile_path
+from util.workers import get_worker_count
 
 logconfig = logfile_path(debug=False)
 
 bind = 'unix:/tmp/gunicorn_verbs.sock'
-workers = max(int(os.getenv("WORKER_COUNT_VERBS", multiprocessing.cpu_count())), 2)
+workers = get_worker_count('verbs', 2, minimum=2, maximum=32)
 pythonpath = '.'
 preload_app = True
 timeout = 2000  # Because sync workers