I'm trying to better understand the differences between Alexa Skill state handlers for audio players. The enumerated states are START_MODE, PLAY_MODE and RESUME_DECISION_MODE. I feel that START_MODE and RESUME_DECISION_MODE are especially poorly documented. For example, I know that a new user coming to my audio player skill for the first time will be in START_MODE. But under what conditions will an existing user (a user that has already used my skill multiple times) return to START_MODE?
Also, under what conditions does my state arrive in RESUME_DECISION_MODE? I can't seem to nail down exactly when this happens. Is it time-based? Is it if another audio player skill has been used in the interim?
It would also be really helpful if a state diagram were published somewhere about these states.