Update sphere_fitting.md

[contsili]

I fixed it with commit 8cfc1e8

[schoffelen]

nice, have you checked?

[contsili]

Yes I checked, and it works correctly.

An interesting insight, if scale=1 then T2*S*T1 = T1*S*T2. In my case (I used my own data) scale was 0.9966 so T2*S*T1 ~ T1*S*T2