deviceDir is used for specifying mount/device host paths, and those
should be absolute paths.

Also, LXC requires target paths to be relative.  Container paths in LXC
binds should never be absolute paths, so we strip any preceeding `/`,
even if a user sets one.

Device hook and devices affect computed node class

Add LXC driver

Looking at NewTaskRunner I'm unsure whether TaskRunner.TaskResources
(from which req.TaskResources is set) is intended to be nil at times or
if the TODO in NewTaskRunner is intended to ensure it is always non-nil.

The old approach was incomplete. Hook env vars are now:

 * persisted and restored between agent restarts
 * deterministic (LWW if 2 hooks set the same key)

This PR introduces a device hook that retrieves the device mount
information for an allocation. It also updates the computed node class
computation to take into account devices.

TODO Fix the task runner unit test. The environment variable is being
lost even though it is being properly set in the prestart hook.

Fix deadlock in previous alloc watcher by emitting last alloc update

drivers/rkt: fix invalid volumes

taskrunner: Return the restart delay correctly

Co-Authored-By: schmichael <michael.schurter@gmail.com>

This is a defensive fast-path as 7c6aa0be already fixed the deadlock.

Fixes a deadlock where the allocwatcher would block forever waiting for
an update from a terminal alloc.

Made the broadcaster easier to debug as well.

Support graceful shutdowns in agent

* add vault integration guide in guides section and move current vault integration content to docs section

* complete guide with image

* fix typos

* rename step 6 and fix typos

* fix typos and awkward phrasing along with links

* fix duplicated step #

* fix typo

* fix links so that pages that pointed to the original vault integration content still point there

This affects exec/rawexec drivers

The group utility struct does not support asynchronously launched
goroutines (goroutines-inside-of-goroutines), so switch those uses to a
normal go call.

This means watchNodeUpdates and watchNodeEvents may not be shutdown when
Shutdown() exits. During nomad agent shutdown this does not matter.

During tests this means a test may leak those goroutines or be unable to
know when those goroutines have exited.

Since there's no runtime impact and these goroutines do not affect alloc
state syncing it seems ok to risk leaking them.

Arguably this makes task.WaitCh() useless, but I think exposing a wait
chan from TaskRunners is a generically useful API.

* move current telemetry page to docs

* fix broken links to telemetry doc

use this for volume names in drivers/rkt to address #1150

Fix linter job

Add support for docker storage options