"macroheating" of the cell, i.e. heating while ignoring spatially small effects like hotspotting, is the easy part, as the constant nature of emissivity regardless of load is easily observed: both open-circuit and short-circuit conditions give roughly the same heating and hence directly heating-related aging. Only connecting a load (in absence of MPPT tracking, even just something equaling roughly the I_mpp at V_mpp, is an improvement) can reduce the temperature of the cell. This is all obvious from basic conservation of energy, so pretty much common sense to everyone except nctnico.
Now the macroscopic temperature probably isn't the only aging mechanism of the cell, and what else happens requires serious understanding of the detailed cell physics apparently no one here has any idea about, me included.
It's not a surprise this is not discussed in cell datasheets. It makes little sense to invest money in this expensive stuff and yet more money to install it, only to not use it. And short periods of non-usage would be irrelevant anyway, clearly the cell does not age while open-circuit for weeks or months, so clearly aging does not increase by orders of magnitude, compared to maximum power point load situation. If aging rate increases by 30%, 50% or even 100%, this does not matter to the manufacturer as the panels are not intended to sit unused for years. 1000% increase would be something that needs to be discussed in datasheets/instructions, and would be a problem when inverters in some situations have to cut production e.g. due to voltage or frequency rise, or utility demand control signals.