Coping with Cross Provider Machine Node Definitions

We need to be able to eventually cope with multiple service providers within the Orchestrator.

One way to do this is to, with the help of other teams, such as Professional Services, plan out the following ideas:

• What constitutes a nano, small, medium, or large machine in terms of GitLab installations?
• Do these sizes change per service node definition?

This would allow us to give customers the option to define things like "Medium" or "Large" based on their user count needs and then, in turn, the terraform will cross apply the right thing per our recommendations.

This allows us to define under the hood what an appropriate node looks like on each provider and come up with some estimations of what it costs to run GitLab in specific providers, or spread across a hybrid cloud.

Logging this as an issue to tackle, this came up during discussion with @ghickey of !1 (merged)

cc @mendeni @ljlane