The twin engine attackers/bombers are the ones that did the real difference in the war, with all due respect to the 4-engine ones. Never got the proper credit for it because tactical bombing, support and anti-shipping are not as glamorous as a mass-scale air war at 30,000 feet. Mosquito bombers did some strategic bombing - if you can call busting city blocks "strategy".
B-25s, A-20s, Beaufighters and Mosquitoes are the unsang heroes. Probably the TU-2 also on that list, except that it is so forgotten (and Russian) that I know little of its operational history.
The big planes are much tougher, but due to the way AH models damage it doesn't feel that way. What is a single section of damage tracking on an Me410/Mossie/P-38/Tu-2 is about the size of three or four sections on a Yak-3. That means the Yak spreads the damage out over multiple damage sections while the big planes get drilled.
The damage model has always been AH's weakest point. Since it is simplistic, deciding how much hit-points a component should have is a bit of a voodoo. It is not like the engine and aerodynamics parameters that are documented and have test flight records. In the case of the Yak3 it is clearly way off the mark, turning a plane that sacrificed armor and metal parts in favor of light weight and reduced costs, into one of the toughest planes in the game. Not that wood is weak, as the mossie demonstrated, but there is a difference between replacing some metal parts with wood vs. basing your whole design around an optimized wooden construction.
The problem is that a great damage model does not bring many new subscribers. A new terrain engine might, so HTC got clear (and correct) priorities. I'd love to see some small improvements to the current system, such as:
1. bomber gunners getting wounded and killed - and when they do, they don't keep showing tracers coming out of their guns...
2. damage that add drag to the airframe. (modifying the lift is perhaps too complicated). Tearing big cannon holes into the airframe, or turning it into a sieve with 0.5s cannot be good for the airflow...
3. some scaling of the hitpoints by the size of the component to reduce the "Yak effect" a little.