I don't have much faith in leak down tests based on what I've seen, its just another number really but does give you some insight.
I would not try to hold or move an engine with 100 psi in the cylinder, it is pretty difficult in my experience. In fact my car with a 4 speed and 11 inch drums, adjusted very well, will eventually push through the E brake with 100 PSI in any cylinder and the trans in gear.
If you get the cylinder you're testing at TDC it will stay there when its pressurized. I really don't see how a leak down test anywhere other than TDC is giving you useful info.
The only time I have ever gotten "hero" 2-4% leak down results were on engines that were in the 120-130 PSI range when perfect would be 160. They had worn rings, and were at the bottom of the compression test spec, but they were small engines with relatively tight ring gaps. If you gap your rings at .024 I don't think ever going to see results like that. I'm starting to think that the results are easily influenced by ring gap and bore size.
My personal opinion is that anything below 15% is perfectly acceptable, 10% is good. If you've got 20% or more its high, and it will show up in a compression test, but you will still make power and the plugs will look OK. There is other info out there that will certainly disagree. A fresh engine with decent machine work really should be around 10% or less though.
your experience agrees with mine. For many years the OEM said up to 40% was OK, a few years ago they changed that to 20% as being the acceptable limit. I seen a guy almost lose a finger by trying to hold an engine with a breaker bar when the air was hooked up, could not hold it, and got between a stainless steel fan blade. Zero gap rings will fool you as well when doing a leak test. They always test good unless you got bad damage. Engine started filling vapor tank with an air pump every run, leak tested good with zero gap rings, but in pulling engine apart, rings were just worn out.
