alert if a host runs a puppet environment other than production for too long

with the puppet prometheus report processor, we're getting metrics that tell us what environment was last reported as being used on a machine.

we can now attach a puppet agent to an environment that's not production and it'll make it possible to run puppet code that's in a branch of the same name in the control repository. that's an important tool for development.

if we have hosts that stay glued to an environment that's not production for too long, it has a good chance of lagging behind and not getting applied some important changes that we push to production, so we should know if that happens.

the alert should be at info level since it's not an operational problem that needs fast attention from the person with star role.

the length threshold for a machine running non-production envs is not yet determined. maybe 2 weeks?