Ping managed devices to check whether they're still alive

Implements #39 (closed).

As pinging all devices successfully only takes tens of ms, I completely dropped the discoverManagedDevives slot and perform the ping before every pipeline instantation. The 2s timeout was suggested by @flucke to be robust enough, the framework only waits 1s to check for name collisions.

While I was at it, I simplified the entire managed devices updating.

@degon @hammer