Reducing the Latency and Area Cost of Core Swapping through Shared Helper Engines