You've pretty much got it nailed. Not many people actually figure it out.
There are a few considerations, some of which are modeled in AWS and others that are observed moreso in real life:
1. Runway length - in general, the shorter variants use less runway, so this is helpful for hot and high operations as well as well, short runways.
2. Shorter variants have lower crew requirements
3. Shorter variants tend to have lower landing costs (as landing fees in real life are a function of either MLW or MTOW). I'm not sure how they're computed in AWS, so it may or may not have an impact.
4. The 736 is a turkey of a performer compared to the rest of the family and even the plane it succeeded, the 735, as the NG frame is significantly heavier than the classics.
5. There have been baby busses (A318s, the 736 equivilant) being scrapped in the real world that are just a few years old because the sum of the parts was worth more than the plane whole.
6. Usually the shorter variants have longer range. In the real world, the A320 and A321 will struggle to make a US transcon flight westbound during the winter without a fuelstop. The 319 can do it without issue, as can the 737NGs.
Now, with regards to your question about running 3x A321's on a 300 pax route: thats just stupid when you could be flying 2x on that route and 1x on another that has unmet demand :-).