Crazyjr: The bad part about running the BEC that hard is the it sin't even attached to the heatsink. It's hard to tell, but it looks like the regulator for the MM is actually on the other side of the board. So, the heat is being leeched away using the PCB traces and into the FETs, and then into the heatsink. Not exactly a direct connection and that heat is just making the FETs not run as well since FET current capability gets derated with higher heat.
AAngel: Sort of. The difference in efficiency between linear and switching regulators becomes very close when the supply voltage is closer to the output voltage. It almost becomes a wash. As a matter of fact, if the input voltage is very close to the output voltage, a linear regulator usually has a lower DO (drop out) voltage and can become
more efficient. Weird eh... :) I'm sure switching regulators could be built with lower DO voltages, but designers realize these will mostly be used with higher pack votlages, so why bother? Linear regulators work just fine for low voltages.
To get an idea using real numbers, play around with the calculator at the bottom of
this page. The switching BEC in the calculations assumes an efficiency of 85%, but that's usually at a decent current draw and higher voltage. When the input votlage is close to the output voltage, and current draw is either quite low or quite high (near the UBEC's limit), the efficiency drops to about 75%.