[FieldTrip] fixed bug in MEG localspheres forward model

[Robert Oostenveld]

Dear FieldTrip users,

We found (and fixed) a bug in the MEG localspheres forward model, which may have affected your recent analyses. The consequence is that MEG source reconstructions would have been of very poor quaility. This was pointed out to me by Fraincois Tadel (thanks!) who noticed that dipole fits of clean topographies ended up in completely unexpected locations. Also MNE and beamformer results will have been wrong, although there it was less obvious to detect the flaw. 

I have included the fix in SVN revision 11061, which you can also find on https://github.com/fieldtrip/fieldtrip. If you use the download version from our FTP server: the version that is released this evening will include the fix.

Further details can be found on http://bugzilla.fieldtriptoolbox.org/show_bug.cgi?id=3037

Sorry for the inconvenience,
Robert

PS If you use singlesphere, BEM or singleshell models, you have not been affected by this.