[TODO] Implement a way to make a MIDI 2 device unavailable to clients until all discovery has been completed

Upon connection, our service logic starts with the Group Terminal Blocks to create the MIDI 1 ports for a device. This way, because Function Blocks are optional, we have usable MIDI 1.0 ports.

Then we initiate discovery and protocol negotiation. After several seconds, when we receive all function blocks from the device, we recreate MIDI 1.0 ports based on the new names and enabled/disabled state of the function blocks.

This results in a period of 5-20 seconds, depending on the device, during which the ports for the device churn.

It would require rethinking how we do discovery and protocol negotiation in the service, but to avoid this churn, it would be beneficial to not show the device or its ports to apps until after negotiation/discovery has been completed. There will still be some churn later if the function blocks are updated, but that typically happens because of actions the customer took on the device.

Related #816