Andrew,
I'm not sure what you mean by flushing __pycache__ directory. So first I deleted the .pyc files in the main __pycache__ directory (no difference). Then I deleted the .pyc files in the components/device_tracker folder (still no difference). I stopped HA, deleted the files, then restarted.
Ryan,
I wasn't getting anything to the log file. I'm not sure how to change the log level (correctly), so I just went into bootstrap and changed it to .INFO. Unfortunately, the errors aren't written to the log, only the console. As near as I can tell, everything is working fine with the tracker, it sees my devices and updates them appropriately when I leave and return.
What I noticed is that the nmap does it's initialization without error:
INFO:homeassistant.loader:Loaded device_tracker.nmap_tracker from homeassistant.components.device_tracker.nmap_tracker
INFO:homeassistant.components.device_tracker.nmap_tracker:Scanning
INFO:homeassistant.components.device_tracker.nmap_tracker:nmap scan successful
INFO:homeassistant.components.device_tracker.nmap_tracker:nmap scanner initialized
< several state settings of my devices snipped>
INFO:homeassistant.components.device_tracker:Device tracker interval second=range(0, 60, 12)
INFO:homeassistant:Bus:Handling <Event service_registered[L]: domain=device_tracker, service=reload_devices_csv>
But when it comes up a second time (and all subsequent times), then it has the error:
INFO:homeassistant.components.device_tracker.nmap_tracker:Scanning