watch_* routines have issues with ctrl-c on linux if no data is received

When we block on `HardwareManager.iter_reports(block=True)`, it is platform dependent whether a Ctrl-C will break us out of that lock acquisition. It looks like on linux the answer is no since it's stuck in the `syscall` on the mutex forever in the kernel. We'll need to implement some kind of internal timeout that checks for ctrl-C and raises, otherwise blocks again rather than doing a true blocking `queue.get()` to that Ctrl-C works on all platforms.

See https://stackoverflow.com/questions/36817050/interrupting-a-queue-get