Add a telegraf interface for metrics

[Barry Price]

It would be nice if we could see, probably via grafana, metrics on livepatch for each instance in any given environment, e.g.:

1. Rundown of units by running kernel
2. Time since last refresh (if the client exposes this)
3. Value of checkState, per unit
4. Value of patchState, per unit
5. Maybe uptime?

There's probably a bit of overlap here, I need to look more closely at what useful info the client actually exposes. But still, this will do to start with.

Writing values into a file that telegraf can read from using its 'file' interface is probably the simplest approach for now.

[David Lawson (deej)]

You may want to think about something that can be run via the inputs.exec plugin, that's super trivial to do, I knocked out some for the retracers in an hour or so a while back. The telegraf plugin story (i.e. plugins requiring compilation) seems kind of unfortunate but it looks like you may be able to dump a config into /etc/telegraf/telegraf.d/whatever.conf with your inputs.exec config and have it work sanely, I'll have to test that out.

[Haw Loeung (hloeung)]

Could also have a little script invoked from cron that pushes these via influx.

[Haw Loeung (hloeung)]

canonical-livepatch is no more and replaced by ubuntu-advantage.