That is after the fact info. What I need is the conditions on what happens in the transitions between these mode. By the time you see the change in the lights, your bird is already crashed.
When I did my test, you have about 1 to 2 seconds to recover.
With the experiment that I have done, in the P-mode it is definitely GPS priority over VPS even at close proximity to the ground. Then RTH interactions comes into play and screw everything up.
There are just too many combinations and permutations to try. I do not mind trying and experiment, just that with limited supply of parts (mainly props), I do not want to do too much playing around and risk not able to fly the P3 completely. I guess one can always use the P2 props and some tape on the threads to do testing.