bind operation (in transports) could be asynchronous

[gdamore]

Right now we insist that bind is fully synchronous. The idea here is that bind(2) is actually synchronous. The reason to support making it asynch is so that we could react to different IP addresses on the host -- meaning we could run this in a background and retry binding.

This would more closely match what nanomsg does, but to be honest this behavior would be contrary to typical application design.