I was told it is the equivalent to the energy that you get from a candle that is positioned a hundred miles away!).
Probably true, but irrelevant, because the signal has predictable properties which enable its recovery from below the noise floor (
Shannon, 1949 or something like that).
The GPS signal recovery is no less reliable than any other radio link. Look at mobile phones - much greater field strength and a lot of calls are still unworkable. My experience of GPS is that ~ 99.999% of the time it is rock solid. The altitude solution (KLN94 or Garmin 496) is usually within 20ft of a known elevation, or altimeter reading at a low level and with a known accurate QNH.
I would guess that the regular signal losses described here are due to poorly installed units. Most avionics shops do not do a decent job, and inter-equipment interference is very common.