Problem there is scaling health. Objects' health maximums are set in stone, so reducing health would be pointless as people could repair it anyway, and would be detrimental to an attack team if the player count changed during their attack anyway. Oh, you got that NY down to 100/750 health? Too bad, someone left in anger and now the playercount says it should be 650 health, so it's back up to 650/750 health now. Who needs golden wrenches when you can just ragequit?
And I get the feeling swapping out the entire building controller for a new one will not fly (last time I remember spawning a building controller midgame it led to a crash), and even if it did, it would fully heal it every time the player count changed anyway. There's a reason the defense logic is handled by replacing the armour class with a different one. Which you can't do with actual buildings because buildings have two armour types: one for the exterior and one for the MCT. I haven't tried using the armour replace script on it but there are clearly 4 things that could happen, ranging from undesirable to nothing:
it replaces the exterior armour and not the MCT (meaning repairs and infantry assaults are equally effective at all player counts)
it replaces the MCT armour but not the exterior (so it only affects repairs and infantry assaults, and not naval/air rushes)
it replaces both armour types with the same new armour type (so the MCT and walls behave identically, either making repairs/infantry assaults completely ineffectual if we choose the exterior armour, or making repairs too easy because a golden wrench works from anywhere if we choose the MCT armour)
it doesn't replace any armour (so it just doesn't work)
Also, Allies already win too easily by Longbow rushing the sub pen. Scaling its health down as it becomes easier to amass the 5 longbows required for a kill sounds... counterintuitive? Same goes for armour, though to a lesser extent if it follows the same "don't affect limited-ammo units" rule as defenses - but if the LBs fail, the rocket soldier pilots will just have an even easier time finishing the job.
Maybe the "rushes too easy to spot" issue can be fixed by fog being even thicker than it already is - but then Longbows become even more overpowered at base destruction - something that is, again, normally not an huge issue (but still something I'll be nerfing with slightly lower building damage) as on other maps it's not too hard to spot an LB rush before it happens and then either intercept with rockets/TTs or just engy-camp the expected building - the latter being the only option here. Maybe the sub pen needs a set of battlements made out of indestructible SAM sites, devised by the Soviets after seeing how incredibly effective the castle is on Siege?