How well does testing a players reaction speed work for making the game hard?

Say the server is perfect and predicts where the player is when the AoE cast finishes so the 0.3s update does not matter anymore.

You're playing with 50ms latency. Human reaction time is about 215ms.

The AoE is cast at the players feet and is big enough that it takes 0.700s to reach safety after you start moving.

TIME
0.000 Server starts casting the AoE
0.050 Client receives cast beginning and shows red AoE indicator
0.265 Human reaction time - the player begins to move now
...
0.965 Player leaves AoE area on the client
1.000 Server finishes cast and checks predicted position - player is in the clear
1.050 Client receives cast finish and shows the player is OK

Now what happens if we have 150ms latency

TIME
0.000 Server starts casting the AoE
0.150 Client receives cast beginning and shows red AoE indicator
0.365 Human reaction time - the player begins to move now
...
1.000 Server finishes cast and checks predicted position - player is in the AoE, so it's a hit
1.065 Player leaves AoE area on the client
1.150 Client receives cast finish and shows the player taking the damage

There is the problem. If the game is tuned to make it hard for a player with 50ms latency it is impossible for the same player with higher latency!

So using reaction time to measure a players skill can never work unless all the players have the same latency.