Didn't Need An Army

[Aquillion]

I mean you don't need an army, obviously; given the wide range of powers and options in D&D there's always another way. But you have to get past the city's defenses somehow, and that's a huge deal. That said, I think that that undersells what the most important part of Soon's defenses was (the one that nearly killed Xykon) and why it was so effective.

Anyone who tried to bypass the traditional defenses risked getting ganked by the Ghost Paladins. And almost nobody knew about the ghost paladins! You carefully structured your plan so the villains "coincidentally" discover them, but that would have to be total luck, because they were a massive secret and did, in fact, manage to take the villains by surprise.

Your plan also requires massive amounts of luck and effort at various points; there's numerous ways it could go awry. Could it work? Yeah, sure. No defense is going to hold up if the villains roll natural 20s at every key moment and the defenders fail at every opportunity they have to stop them.

Most of the other gates had one or two clear single points of failure. Durkon and Lirian's gates relied on a single big powerful spellcaster, who generally had a few narrow tricks, and were relatively vulnerable once those were eliminated. It's hard to evaluate Girard's gate because we never saw its defenses working properly, but it seems to have been all-in on illusion. On top of this, all of their defenses (ironically including Girard, who prided himself on trickery and deception) were basically what you would expect if you knew anything about the people in question - part of the reason Soon's final trap was so effective, IMHO, is because nobody would expect a Paladin to do something like that.

Looking at your plan, there are numerous high-risk points of failure (I would rate the chance of each of these steps succeeding, individually, at less than 50%). Also, as a key point, since you rely on them not knowing they're being infiltrated early on, they can't actually advance with their army - if Kubota is discovered to be collaborating with goblins and a lich at any point before he's in full control, he immediately ends up in jail at best and likely executed.

• They're extremely likely to be caught in the scouting phase, before anything else, at which point the entire plan basically goes down the drain. The Sapphire Guard, after all, has multiple spellcasters, including a high-level seer; and the entire place is filled with Paladins who can detect evil at will and who are generally relatively incorruptible. Losing at this point not only ruins your plan but also ruins the surprise of the military siege backup plan.
• There is no guarantee that they successfully make contact with the right nobles. Just because someone wants power doesn't mean they're stupid; it doesn't take much to realize that Xykon and Redcloak's long-term plans are incompatible with humans retaining any sort of power in the city. Therefore, it's extremely likely that whichever noble they contact goes straight to the Paladins, at which point the plan goes down the drain.
• Even if Kubota plays along, framing Shojo is hard. Again, as a reminder, the Paladins can detect evil and the city employs multiple high-level clerics and wizards, including a Seer. And getting into power is even harder. If Kubota could do all this, why hasn't he done it already? What are Redcloak and Xykon actually offering him that he doesn't have access to already? (Remember, neither can actually enter the city while it's under the control of Paladins, so they can't provide direct magical support - at least not without massive risks which they have no reason to take.) I feel like you're massively overestimating Kubota's competence here - he clearly thought he could pull amazing stuff off but in reality his track record doesn't suggest he's actually a big player.
• But it could get even worse. Kubota is a backstabbing treacherous snake. He knows that Redcloak and Xykon's plans are completely incompatible with his own; therefore, the most likely thing he does when they contact him is pretend to play along, milk them for as many resources as he can without actually doing anything that would get them into the city, then backstab them at the most crucial moment in order to look like a hero. Maybe he fails at this? But there's a high chance that Kubota plays along until he's in charge, then immediately betrays Xykon and never disbands the Sapphire Guard; or, worse, pretends to disband the Sapphire Guard in order to lure Redcloak and Xykon into a trap. Why on earth would he give them the city when he doesn't have to? He probably doesn't like Paladins, but Evil Is Not One Big Happy Family; the Paladins have to follow the law and are generally favorable to humans, so at the end of the day he's going to choose them over Redcloak and Xykon.

Honestly the last point is the big one. What possible reason would Kubota have to actually give Redcloak and Xykon access to the throne room? Why would he even consider doing that? At that point he has what they want and they have no leverage to make him do anything (no, a Lich and a goblin saying "he's totes evil tho" isn't going to mean anything to him if he already made it through all the deceptions needed to get to this point. Even if they do, the Sapphire Guard is not stupid and will prioritize eliminating Redcloak and Xykon over him.)

He's not going to submit to any sort of magically-binding contract unless they have something to offer him that would be worth the vast risk that implies, which they simply don't; and even if they did, he has access to a bunch of wealth and could probably just pay a spellcaster to undo it. Even if everything else goes properly, any human noble they work with is going to immediately want Xykon and the bearer of the Crimson Mantle dead once they're in power in order to clean up loose ends.

Compare this to the other gates, where the plan is generally something like:

• Win one big fight.
• Overcome one big magical working.

...both of which are much more doable. Honestly even the normal way to get through Soon's gate is similar, it's just that the "big fight" is on a bigger scale. Your plan adds a ton of complexity and points of failure for nothing.