Consider less implementation-defined definitions for PressureState

[bvandersloot-mozilla]

Hello, coming over from Mozilla's standards-position thread on this spec. I have feedback around one concern.

The map from contributing factors (and those contributing factors themselves) to PressureStates are left very implementation-defined. This introduces risk that an application that relies on this would behave very differently between browsers. Consider a webapp developed against the heuristics employed by one browser, running on a browser that provides less aggressive signalling of pressure. The webapp may not take the warnings on the second browser for the dire warning the are and keep pushing for a bit. This would lead to some hard-to-diagnose and vague webcompat bugs of intermittent reduced quality under some conditions for one browser for that app.

I'm not familiar with the OS signals exposed, or how each OS decides when to throttle processes, so I don't have too much to go on here. However, did consideration go into specifying the production of the PressureStates any more closely?