A bit late, but I don't agree with the point that there are not a lot of subhourly data. Today (and it was the case last year when this was posted) almost all satellite data providers have subhourly data (stallite images are now 5-30min).. One might even say that a lot of providers were forced to adapt to this limitation of PVsyst and provide hourly files...
A true simulation on a lower timestep would be a huge gain in precision / effort on our end...
And naively, I don't understand the difficulty... a part from the thermal model that has inertia, all the other elements can be establish at whatever time step... I agree that the computing time would probably be 4 fold if one let us say wants to have 15min simulations... but I don't think this is a real problem.