wildfly-server/wildfly-host-controller currently depend on 2 subsystems: remoting and io.

This is problematic for a number of reasons:

• It prevents either module from leveraging wildfly-subsystem (this would create a circular dependency)
• It prevents either module from creating deployment unit processors

As far as I can tell, there is no technical reason for this, other than poor encapsulation.

To address this, both remoting and io should have their own SPI module, that includes interfaces, service descriptors, attribute definition implementations, etc. needed by wildfly-server/wildfly-host-controller and any consuming subsystems.