This is an interesting one, so I tried to find a bit of data on that. And from reading what they are saying, the MPPT will drop more from it's peak output when it gets hot compared to a PWM from it's peak. But here is the problem with that argument. The peak power of the MPPT is quite a bit higher, so as it falls off in the heat, it just comes down and reduces it's lead a bit over the PWM. But I can not find any situation where the PWM will out perform an MPPT in actual watts out. The one good argument for a PWM is with the cost, especially on a very small system. For the cost difference, you could buy another solar panel, and then the PWM system with 3 panels does make more power than an MPPT system with only 2 panels that cost about the same. So sure, if you have the room for another panel, the PWM could win, but if you are limited to a given amount of solar panel, the MPPT will make better use of the panels and extract more power.
While it is true that at a small difference in voltage, the MPPT can't step up the current, it will still virtually match the PWM. And with the panels in series, the voltage will go above the battery with less light on the panels than if they were in parallel.
An MPPT controller does have a processor inside to run an algorithm to track the voltage and current and make constant adjustments to optimize the power conversion. Just like any small computing device, that processor will use some of the power you are collecting. So in a very low power situation, the energy needed to run that processor may become a fair part of the load. So obviously, for a 3 watt solar cell running a security light, an MPPT control would be silly.
So it does always come back to how much power do you need? How much space do you have? What is the budget?
If you have the room and want to save money, a PWM may work well. If you need the most power in a limited space, and the budget can handle it, an MPPT will make more power.