fsim is a simulation which runs a trial X number of times, there's some random variance in it. Once you hit min delay the increase is pretty small and was apparently lost in the noise...
Also while blade hands will increase your damage over claws, try an fsim of statue form if you want to see it really beating GSC's
edit: Got unlazy, ran a few.
Because ctrl+F fsim isn't working for me, here's some single results from f with fighting, unarmed, and maces and flails all at 27:
20 str, 10 dex, vs stone giant
unarmed: 102.0 average hit, 312 max hit, 7.5 auts, effective damage: 131.3
GSC: 108.4 average hit, 461 max hit, 10.5 auts, effective damage: 96.7
30 str, 20 dex, vs stone giant (same order as above, auts omitted, it doesn't change)
unarmed: 149.5, 523, 190.5
GSC: 136.5, 618, 122.4
40 str 30 dex vs stone giant (go go chei powers)
unarmed: 188.1, 684, 240.4
GSC: 161.6, 830, 145.1
So when going with chei, and using statue form as you properly should, claws can be around 55% more (30 str) to 65% more (40 str) than a +9 freezing GSC. Also, since you're slowed in statue form, claws attacking at 7.5 auts means you won't see double actions from monsters (unless it's something very, very fast). Shields would remove the offhand hit, but these tests were done in a melded robe/no shield.
Bonus: Necromutation, 40/30 str/dex:
unarmed 95.2, 371, 184.0. Not as good as statue but hey, decent damage still. I'd do 152.4 in base troll form, so I assume most of that damage is the draining brand, which wouldn't work in extended anyways. Ah well. I am still playing that troll lich...