A postal scale is better than a fish scale, either one off the fish or one to weigh him with. 
Digital trigger scale with a 1 oz. minimum may be ideal, but not cheap.
Cars that weigh less than the NMRA recommendation will allow you to pull more, IF they function well in all modes of use. Only you can make that decision. Two cars that are for all intents are ‘identical’ may not weigh the same to get excellent performance, and/or they may not roll the same. Therefore ‘tonnage’ may not be the best indicator.
My Y3s (well, the four I have tweaked, anyway-no traction tires on any darn it) will pull 30+ hopper trains up a 2% grade, no problem. My P2K Berk will do 24. My Athearn Mikes will do 21, 17, 16, and 16 respectively. Everything else will do more, freight wise. My Athearn Pacific will do at least 3 passenger cars - I don’t know what it will do maxed out because that all it ever has to pull. Everything else will do more - BLI Class A 75 plus something - ran out of room. The plating wearing off the tires a little on most any engine will help a lot, IMHO. Personally I love traction tires, but I am in the distinct minority. I do not see what prototypical has to do with it when talking about a steam loco powered by an electric motor, but again, that’s just me. When in doubt, add weight.
Unless you have 50 locos or more, the best test may be to find the worst and best pullers and then rank the rest relative to them. Doubleheading is fine and fun, but two mallets on a 30 hopper train ain’t prototypical either. When you get a new loco back it up to the best and see which will pull which for starters.
Nice looking layout by the way.