Personally I think the Covert Action rule is about focus, not the sheer number of different component games. The rule kicks in only if the components are complex (or otherwise confusing) enough to distract the player from each other. This is something that can be seen in the Civilization series - there's a lot of stuff, but it's simple enough not to distract the gameplay.
I guess games that adhere to the KISS principle aren't likely to violate the Covert Action Rule anyway.
Forgive me for debating this point with such enthusiasm, but, obviously, I am pretty passionate about game design.
Civilization is a good example of the principle you're giving, but it's not comparable to the games I'm talking about. Civilization is one game with a lot of features; diplomacy, city building, things like that. You could loosely call those minigames, but they're more like... nodes of complexity. Any of them, divorced from the whole, would be unable to stand on its own. There are a lot of mechanics and features, but there's only one real game. Star Control 2 is another game along the lines of what I'm talking about; it takes the action battles of Star Control and appends a huge and ambitious adventure/strategy campaign, changing the genre of the sequel by appending a completely different and much larger game to their battle system. Star Control 2's combat "minigame" literally stood on its own -- it's the original Star Control grafted onto a completely new genre of game. Yet the game works, it's a beloved classic. These examples I'm talking about have multiple radically different modes of play. They're not just full of features, or even minigames.
Space Rangers 2 is a great example of a game where the sub-games are sufficiently complex to distract from the rest of the game, but it works despite and because of that depth. I've poured hours into trying to win RTS battles, or completely forgotten about the rest of the game while chewing through text quests, in both cases finally finishing the mission only to be reminded that I just got a reward equivalent to five minutes of gameplay (and that doesn't bother me, because the real reward was the fun of playing the side game). It's a space trading and combat RPG, but they have missions about being in jail, running for president of a planet, and even being a small animal and trying to get fat and popular while digging a big burrow, each a game in itself with its own carefully considered mechanics and balance. Multiple games in one? Oh you betcha yes. Simplicity definitely isn't something I would associate with the game. The Total War and King Arthur games both combine two mechanically deep forms of gameplay, switching between the modes and spending a long time in each one. It's as if civilization took the time to spend twenty minutes simulating a real-time battle every time your armies clashed with a computer. Some strategy games have combat minigames, but the tactical combat in Total War games isn't a minigame, it's half the game. Arguably the more important half. But neither is the strategic layer a mere minigame; they're inseparable because in one you build your armies and decide to care about the outcomes of battles, and in the other you put your armies to the test and are tactically challenged to achieve your strategic objectives in a moment of truth.
I don't think simplicity is the key; it might keep a game from running afoul of the rule, but there are complex games that openly break the rule and get away with it anyway. A game like X-COM works because everything matters. Every rocket costs money, every death is a character whose experience is lost, every alien weapon dropped to the floor is an artifact to be reverse engineered, every bit of collateral damage to the spacecraft is less scrap for your engineers. And the relationship works the other way too, with every little strategic development impacting your tactical play. Total War and King Arthur use the same approach, though they have less feedback from the tactical game back into the strategic game. Space Rangers, in contrast, is highly diffuse and barely integrated at all; they make it work through pacing and Elite-like freedom of action. You pursue whatever game mode you feel like at any given time, skip some if you deplore them, or give up and go back to another mode if you're finding one too difficult. Don't like the RTS game? Don't accept missions of that type. You can even tell the game to never offer them because you refuse to ever take them. The component games are barely connected, but you're never pushed around, forced to do something you don't want to. Instead, you choose to change the pace every once in awhile, and it works remarkably well. These can be contrasted with Covert Action, which, if I recall correctly, interacts its parts only through success and failure. I think you can avoid the infiltration game sometimes by using alternate methods to gain information, but it's an integral part of your toolbox for solving problems, and sometimes you need it. That creates a pacing problem. Gangsters is more integrated, since the action sequences are watching your strategic orders carried out and giving new orders if you need to change them on the fly, but there's generally little for you to do. It's a major lull in your interesting decisions, mostly about spectacle, and the spectacle isn't that great.
I do recognize that there are games that suffer for multiple game modes, but this is one of the game design principles that I think is too conservative. Following it would save games like Spore, but it would also kill off games that actually thrive while hot glue gunning multiple genres together and cackling like a mad scientist. Sometimes I do get tired of one of the two game modes in the Total War series, but it's still far superior with them combined than if they were split. The two games just have tremendous synergy, since I care much more about everything in both modes, since they're key to success in the other. Likewise, X-COM is beautiful in part because of how integrated and multigenre it is.
I don't think these games are exceptions. And if you look for them, it's striking how many games are already heavily modal, swapping between subgames. Even LCS's basemode and sitemode are essentially two different games, each able to stand on its own. Oubliette was only one of those games; LCS evolved a second game on top of it with its tumorous feature creep. But it works, because they draw on the same characters, have heavy feedback between them, and that's enough.
I think amalgamating multiple games isn't inherently a bad idea. It's a deceptively cool idea that is very dangerous from a developmental standpoint. But it can work, sometimes spectacularly. It just has to be done well.