Beliving that the difference between the two characters is "50%" and claiming that that, by itself, is fair is a non-sensical argument.
Without staring yourself blind at numbers, look at this:
A human without skill-focus will manage to disable roughly one out of four traps.
A halfling with skill-focus will manage to disable roughly nine out of ten traps.
Two parties enter the same room.
There are 20 of the above trap in it.
The party with the halfling will be subject to two traps and the one with the human to 15.
Numbers:
The human has a 25,9 chance to disable the trap, the halfling has a 89 % to disable the same trap.
Henche, the halfling is 247% more likely to disable the trap.
I consider the disreprancy huge, it should be fixed.
Requiring two consequtive failures, as I described above, does that quite neatly.
In the above room, the party with the human would be subject to eight traps. The one with the halfling to, on an average, 1.5 traps