Fall back to OVSBridge if no controller is available for default switch

This should fix #398 for real; note that if we try to shut down ntpd
right at boot, it doesn't work! ;-(

However, setting the Unix time in the traditional manner using seconds
since 1970 should do the trick!

fix silent failures when rt cannot be assigned - will follow up on this

Conflicts:
	bin/mn
	mininet/node.py

merge HostWithPrivateDirs into Node

ensure we retrieve a single PID when run on a cluster node

Background: the reference controller is reactive and installs exact
match rules. By attempting to start a telnet session we make sure that
the ARP caches and TCP flow rules are set up (in one direction at
least) before the test starts.

This is intended to help with #413

add cli usage information

support wireshark versions greater than 1.11 in test_walkthrough

Mininet crashes when running ping between two hosts with no interfaces

if a test fails or exits with an error, run cleanup as a precaution

Fixes #412

wait for sshd to start in baresshd example

use node.pexec() to update IP address of intf instead of node.cmd()

Unfortunately, this can slow things down a bit - perhaps
100-200 ms in the case of ovs-controller, but I am hoping that
it may help slightly with #399.

update interface IP address with pexec so that backgrounded process output from the cli cannot interfere