Coeus, you have a completely incorrect idea of not only how G&W operates in Brawl, but also how competitive Brawl works as a whole.
First of all, there are no dive kicks in Smash. The Smash engine doesn't allow for overheads or crossups or block stun. The Smash engine doesn't limit a character's jump to a pre-defined arc. The positional advantage dive kicks give in Capcom fighters simply doesn't translate into this game. Diving moves in Smash are no different than regular ones except they force the user to fast fall. This means that it actually LIMITS the potential of a move because the option to float a move in a spot is gone.
Secondly, G&W in competitive Brawl is completely different than the character you described. No character has combo's in Brawl. Any string of damage you get comes from obtaining positional advantage over your opponent. G&W was initially a top tier character in that metagame simply because he has huge disjointed hitboxes on his fishbowl nair and turtle bair control a huge amount of space around him. Those two moves are still the heart of G&W's metagame today and they make him a powerful ZONEING character. The reason why he absolutely cannot be played as a rushdown is because he has basically no fast moves and he struggles to stay safe if his opponent shields his attack. In Brawl there is no such thing as offence, and G&W is no exception.
As for the merits of G&W getting his diving dair back in P:M there are very few. Remember that before G&W dives he actually has an animation where he pulls the key above his head. During this animation he actually floats in the air, which delays him being able to land and L Cancel thus making it ineffective for shield pressure. Outside of that, I've already mentioned that manually fast falling the dair results in a very similar effect, so what it could add would be very limited. The only way I can see G&W's Brawl-style diving dair adding anything to his Project:M iteration would be to completely remove the windup and make it actually dive immediately after the startup, but I don't think that's really in the cards for any future implementation. Though it would be fun to see in action just once.
