Reviewed-By: Mike Stahnke

This patch introduces managing remotely cisco IOS network devices
through ssh or telnet with a puppet type/provider.

This patch allows to manage router/switch interface
with the interface type:

interface {
	"FastEthernet 0/1":
		device_url => "ssh://user:pass@cisco2960.domain.com/",
		mode => trunk,
		encapsulation => dot1q,
		trunk_allowed_vlans => "1-99,200,253",
		description => "to back bone router"
}

It is possible with this patch to set interface:
 * mode (access or trunk)
 * native vlan (only for access mode)
 * speed (auto or a given speed)
 * duplex (auto, half or full)
 * trunk encapsulation
 * allowed trunk vlan
 * ipv4 addresses
 * ipv6 addresses
 * etherchannel membership

The interface name (at least for the cisco provider) can be any
shorthand interface name a switch or router can use.

The device url should conform to:
 * scheme: either telnet or ssh
 * user: can be absent depending on switch/router line config
 * pass: must be present
 * port: optional
 * an optional enable password can be mentioned in the url query string

Ex:
To connect to a switch with a line password and an enable password:
"telnet://:letmein@cisco29224XL.domain.com/?enable=letmeinagain"

To connect to a switch/router through ssh and a privileged user:
"ssh://brice:letmein@cisco1841L.domain.com/"

Note:
This patch only includes a Cisco IOS provider. Also terminology adopted
in the various types are mostly the ones used in Cisco devices.
This patch was tested against:
* (really old) Cisco switch 2924XL with ios 12.0(5)WC10
* Cisco router 1841 with ios 12.4(15)T8
* Cisco router 877 with ios 12.4(11)XJ4
* Cisco switch 2960G with ios 12.2(44)SE

Signed-off-by: Brice Figureau <brice-puppet@daysofwonder.com>

This type needs to be started again from scratch, and I'm not
going to do so for 0.24.5.

In particular, the model for red hat and sunos need to match --
they should both use the device name as the actual name.

Signed-off-by: Luke Kanies <luke@madstop.com>

longer uses the :target parameter (which I'll be removing in
the next commit).

generate, rather than using ParsedFile.  This should fix #777,
and has from what I can tell.

with bringing up alias interfaces.

in #744 which broke the templates.
In the process, I also added test code for the redhat interface
provider and rewrote how parsing worked to make it more testable.