Mutiprocess Keystone Service

Currently Keystone services only spawn green threads to handle auth requests. Due to python's limitation, all green threads are actually being scheduled within one system process, which results in only one processor core can be utilized. When requests concurrency increase, the throughput doesn't increase because requests pile-up and the response time of each request goes-up very quickly. This can be a serious problem since keystone is required for other OpenStack services, such as Nova, Glance, Swift. Therefore, keystone's performance(throughput) is critical to the whole solution's performance and user experience. I'd propose configurable multi-process (workers) keystone services to tackle this issue. Similar implementation has been added to Swift, Glance and proposed to Nova.