Just another opinion:
Sounds like you need to take your time and do some testing. There's 2 way to use the plunk test, so far you have only been using 1 of them. Take and size 50 cases and do a plunk test with them. They all should pass the plunk test in your bbl. If not either re-set/re-adjust your sizing die and redo the 50 piece of brass test.. If that doesn't work you probably have an oversized (too tall) of a shell holder. If all 50 pieces pass the plunk test it's time to move on.
Measure the base of the case then bell the case as you normally would and then measure the base of the case again. Do this to all 50 cases, make sure the base of the case is the same before and after belling. A lee 9mm factory expander plug next to a custom expander plug I made for the 9mm's. I load a .358" cast/coated bullet in all 3 of my 9mm's. As you can see the factory lee expander plug is designed for the shorter jacketed .355" 9mm bullets.
Adjust the 3rd die so it only seats the bullets, no crimp of any kind. After seating the bullet check the oal and then measure the base of the case and the mouth of the un-crimped case. Do this for all 50 cases noting the differences in the oal's. Also try the plunk test with the cases un-crimped, so bbl's are loose enough that the un-crimped brass will still pass the plunk test.
Finely crimp the bullets and do the plunk test with each round.
Doing this will not only test each stage of the reloading process, it will give you a real idea of what your dies are actually doing and the tolerances you are actually keeping with your reloads.
I'm not sure where you are getting the .370"/.371" from???? I just went and measured some cheap cbc ammo (.376") and some federal (.377") and some remington (.376") along with my reloads (.377"). I try to target a 3/1000th's taper on all my semi-auto ammo and straight walled cased pistol ammo when I use a taper crimp. The end result is case mouths that measure .377" with 1 every now and then measuring .378". I use a dial caliber and mics to measure my reloads with. I use test gauges designed specifically for setting these measuring tool to read accurately. Coupled with square gauge blocks.
When I seat the bullets I load I don't set the oal for the "MAX" most accurate with lead height. I set everything 10/1000th's less than the "MAX" oal. If you took your time and measured the difference in the oals of your 50 round test run you will see that the oals can and will very that much. Couple that with what others have said about the coating being different thicknesses on the nose profile of the bullets. Two of my favorite 9mm bullets that I use in 3 different 9mm's. The lee 125gr tl tc and the 125gr mihec fn hp. Both are loaded 10/1000th's less than the "MAX" oal using the plunk test.
While everyone does things differently, this is what I did to setup my dies to reload for the 3 9mm's.
Made sure the sizing die was adjusted correctly
Used the correct expander plug for the bullet being used.
Seat the bullet in 1 step to have a 10/1000th's (+/- 5/1000th's) less than max oal.
Crimp in a separate stage putting a 3/1000th's crimp on the case mouth (.377").
That mihec 125gr hp in a nm 1911, a 10-shot group @50ft with reloads using what's written above.
Playing around testing different hollow base pins/designs in a 150gr hollow based flat nosed bullet for the 9mm. A 10-shot group @ the 50yd line.
Back in the late 80's/early 90's we'd fit 6" 9mm bbl's in our 1911 38super race guns. Always set the reloads up the same way testing each phase of the process checking everything & getting it right & tight. Back then we ran full house loads of bluedot and the lyman 158gr rn cast bullets (free lead/358311). Shot countless 1000's of them, don't shoot 9mm's like I used to but I still shoot 5,000+ rounds of 9mm a year. I can't even remember the last time I've had a failure of any kind.
Just sounds like you need to go back and re-test from the beginning of the reloading process and nail everything down. When you get everything right you will be putting out countless 1000's of rounds with boring regularity.