Bug 1099638

XNIO versions previous to 3.0.10.GA had an automatically scaling thread pool based on `java.util.concurrent.ThreadPoolExecutor`. However the implementation of scaling is inherently racy and can result in the pool going idle even though there are executable tasks pending. As a result, clients or servers appeared to "hang" for some period of time. In the event of a client, this may manifest as "connection timed out" or "operation failed with status: WAITING" exceptions. This is particularly prevalent on Windows. In this release of the product, the thread pool code has been modified to behave the same way as WildFly 8+. This means that the customer will observe the following behavioral changes: * I/O task thread pools (in this case, Remoting and Management task pools) will automatically and immediately grow to their maximum size. * The "core pool size" setting of these pools will have no effect, though it will continue to exist going forward in case the previous behavior can be safely reinstated.