Exceptions should be logged (to standard logfiles)

[stianstr]

The devices and daemons each have their own log file. These contain everything that is logged using log-functions in chains.common.log. But uncaught exceptions are printed to stderr (or stdout?) by Python and hence are not logged to the standard logfiles.

Note however that for the docker-versions, supervisor catches stdout and stderr, and these exceptions can be found in /var/log/chains/supervisor. (For devices the relevant logfile is manager.log).

It would be nice if exceptions too were logged to the standard logfiles.

I can think of two ways to do this:

1. Catch unhandled exceptions somewhere, and log them before letting them propagate. Though, I'm unsure how smooth this will work when dealing with threads.
2. Let chains.common.log output to stdout instead of file, and let it be the daemon-starter's job to pipe til to the relevant file (ie. manager's job for the devices, and supervisor's job for the other daemons).

[chrisaq]

@stianstr didn't you fix this?