juju 2 rc1 does not bubble up LXD errors

[Adam Stokes]

Reproduce:

Running juju 2.0 rc1 with LXD 2.2
Load a custom `juju-default` lxc profile with the following:

```
name: juju-default
config:
  boot.autostart: "true"
  linux.kernel_modules: openvswitch,nbd,ip_tables,ip6_tables,netlink_diag
  security.nesting: "true"
  security.privileged: "true"
description: ""
devices:
  eth0:
    mtu: "9000"
    name: eth0
    nictype: bridged
    parent: lxdbr0
    type: nic
  eth1:
    mtu: "9000"
    name: eth1
    nictype: bridged
    parent: conjureup0
    type: nic
  mem:
    path: /dev/mem
    type: unix-char
  root:
    path: /
    type: disk
  tun:
    path: /dev/net/tun
    type: unix-char
```

Make sure you do not have a bridge named conjureup0.

`juju bootstrap test localhost`

The bootstrap will succeed, then attempt to deploy a basic charm:

`juju deploy ghost`

If you watch `lxc list` it'll attempt to start a new container but then immediately delete it. `juju status` will eventually state that the machine is 'down' but doesn't give any other indication of what the problem is.

For me to debug this I needed to run `lxc monitor` while `juju deploy ghost` in order to see that the container was failing to create because of a non-existent `conjureup0` bridge.

That error is displayed via the LXD logging server and would be helpful if this could make it in `juju status` or shown as an error in `juju debug-log`.

I'm sure this is a corner case but happened because I installed 'conjure-up` debian package and then uninstalled that package which removed the bridge but left the existing juju-default profile in tact.

[Joseph Phillips (manadart)]

Vast changes in LXD versions and the provider since time of report.