what I'm sure you understand is that hp/l is the measuring stick, we may be actually saying the same thing tho it's just that hp/l is easily definable and a practical, useful means of quantifying output. Sure you could perform bench testing with a stack of instruments and determine hp/cfm on each engine but real world? Impossible to achieve.
I guess a better stick might be hp/l/liter of fuel. I guarantee that in the example of 2l engines listed above the higher hp/l engine is also the highest hp/l/liter fuel, better (read, more efficient) design.
Design is totally another animal all together, diesels run what about 40hp/l but with generally more torque/l, gas engines typically 60-120hp/l with all of the upper end being a variable valve time type design and typically closer to square hp/tq. This is why the number is important, given the terribly inefficient design of 90% of the ice the better design typically produces better hp/l ratio, sure it flows more but it also uses more of the energy available from the unit of fuel. You seem to imply that an engine of 60hp/l doesn't use any more fuel/hp than a better design of 120hp/l, that is simply not true, at whatever cfm. Higher hp/l is just generally better design/engineering.
The other 'standard', and perhaps better, indicator of performance is, of course, hp/lb because it more accurately relays the real world, seat of the pants feel. But alas, it is also flawed in your 'scientific' method because there are too many variables to consider which, coincidently, gets us full circle to the purpose of the thread doesn't it?