📖 Short term accelerated compute instance - GPU node scale down

[BrianEllwood]

User Story

As a…engineer
I need/want/expect to…Have the Short term accelerated compute instance - GPU node scale down in a timely fashion.
So that…we do not incur unnecessary cost for an idle node

Value / Purpose

Currently this is a restricted release but it would be good to have this configured correctly before going on GA to avoid unnecessary costs

Proposal

Look at the best method for scaling down the GPU node, currently the vscode scheduler has a parameter MAX_IDLE_TIME of 600 (looks like minutes) before closing down an idle pod

Definition of Done

• GPU node scales down in a timely fashion
• README has been updated
• Documentation has been written / updated
• Another team member has reviewed
• Tests are green

[BrianEllwood]

This ticket came out of the work on this ticket