There were still a few conditions under some OSes that would cause
firewall.py to terminate without cleaning up the firewall settings.  'pkill
sshuttle' was one of them.  Ignore a couple more signals to further ensure a
correct cleanup.

(This only affects sshuttle --firewall, which is a subprocess of the main
sshuttle process.  The firewall is supposed to exit automatically whenever
the client exits, and so far that part seems to work reliably.)

For OS X systems without Python 2.5, runpython.c does not compile.
Use python-config to get the paths for the latest version.

This is required so sshuttle can be run from git repository.

The way the Debian package is created is non-standard, and probably
needs redoing anyway.

(apenwarr: also updates to the matching, latest minimal/do)

Not sure if this will fix anything, but it might stop the problem reported
on some MacOS versions where the firewall doesn't get cleaned up correctly.

If the subprocess was trying to write to its stdout/stderr, its process
would never actually finish because it was blocked waiting for us to read
it, but we were blocked on waitpid().  Instead, use waitpid(WNOHANG) and
continually read from the subprocess (which should be a blocking operation)
until it exits.

I guess we were causing the kernel to syslog on every single packet on
MacOS.  Oops.

There was a problem where trying to bind .v4 and .v6 listeners would set them
to None if there was nothing to bind (if, say, you weren't binding an IPv6
listener).  Later, the code then would try to call a member function of the
listener.  The member function would not do anything if there was no listener,
but trying to dereference None yielded the broken behavior.

address families.