I've managed to create a .MEF that allows me to import minute-avg data (GHI, DHI, T_amb, WindVel) for about a month of utility-scale performance test data. I believe that minute-avg data is averaged into it's corresponding hour and then converted into a typical 8760 .MET file. I really wish you could import Albedo, POA & IPOA but the model is then, supposedly, over-constrained and IPOA is not even an option in the .MEF process. It sure would be awesome if PVsyst could allow the program to do one month of minute-avg data (Input & Output). 1440 [min/hr] * 31 [days/mo] = 44640. Essentially, your 8760 capability would be upgraded to 44640 capability. But then you would probably have to rebuild the entire transposition model
The best thing you can do with PVsyst is manipulate the data with Python (or excruciatingly with Excel) and feed 60 qty .MET files to PVsyst through a batch process, treating minutes like hours. Then treat hours like minutes and get your data back into a stable format. To do this right, you also have to do a time shift of each .MET file to avoid needing to interpolate between hour-data of the transposition model.
Better luck with NREL SAM