One thing I forgot to add, and that people might think about a little, was the impact of drag. Recall that Cl appears in the denom of that sustained turn equation. It'll be the Cl that delivers enough lift at some extreme bank theta max such that you lose no alt. This is where relative drag plays in to making or breaking the fighter. You can express, of course, the Cl as a function of the Cd0 and Cd, right? I.e., Cl = ((Cd-Cd0)/k)^.5. and THAT's ASSUMING WE'RE IN THE LINEAR CD VS CL REGION...
Since sustained turn will go up as Cl required decreases, we want to minimize (Cd-Cd0)/k
Remember that k = 1/(pi*AR*e) so that the denom will grow with AR. Leaving that aside since intuitively pleasing anyway, we can go on. Since Cd0 is a constant and Cd is always going to be greater than Cd0 at positive lift, we can say we want minimal induced drag for the necessary lift, yes? And this is why a Spitfire makes a better fighter than a barn door if both have the same wingloading and powerloading. I.e. you want something with a very healthy d(Cl)/d(Cd) as well as the two aforementioned attributes.
I'd welcome feedback on this since I kind of did it out of my head and could be all wet - but it's making sense and I'm lucid, at least right now.