Constructing PineAPPL grids on hardware accelerators