Min-Idle-Instances vs Min-Instances

[Cesium]

So, I've studied the recommended thread, with explanations from lead engineer Jon McAlister:

https://groups.google.com/d/msg/google-appengine/nRtzGtG9790/hLS16qux_04J 

and Johan wrote:

If you want another behavior like a new slider for Min-Instances please fill a new feature request. 

Has anyone create such a feature request? Where is that puppy so I can star it?

I'd like an explanation of what Jon means when he writes:

 Then, if the non-reserved instances can't keep up, then it will make use of the reserved instances.

Let me re-phrase it, if I may:

Then, if the dynamic instances can't keep up, it will make use of one of the N idle instances.

Could you define "can't keep up"? Under what circumstances does one of the N idle instances serve a request?

Also, where can I find definitions of these terms used to describe instances:

cold

warm

hot

David