TransportTests add queues to bindings which are not necessarily compatible with the transport

[danielmarbach]

The transport tests create input queues and error queues that potentially don't follow the conventions of the transport and add those queues to the queue bindings class. I think we should call BindToLocalEndpoint and ToTransportAddress before we pass those addresses to the queue bindings to better comply with the receive feature.

// cc @andreasohlund @Scooletz

[timbussmann]

if you call BindToLocalEndpoint, the address received from ToTransportAddress might contain transport specific information like schema or machine name. This might fail to create the input queues (although I'm not a ware of a transport doing anything more fancy than adding the machine name. But @SzymonPobiega knows better).

[SzymonPobiega]

@danielmarbach can you give an example? I don't think going through Bind is a correct thing to do for these queues because it is hard to tell which endpoint should bind them. E.g. for MSMQ it is common to have them on different machine, for SQL in different schema. I think the transport should decide how to name them.

[danielmarbach]

@SzymonPobiega that was my point. Currently the transport is not involved in naming those queues. The transport tests you give them a name and add those queues to the bindings which obviously bypasses the transport entirely