I'm just gonna put out there that a game taking a long time to finish isn't the same thing as a game that reveals its rewards over many playthroughs. Of course there's a point when the game stops surprising. But let me make the comparison to a book.
When you read a book you read all the words and when you get to the end you might think about the book. But perhaps going back and rereading it from a different mindset or after experiencing some other insight will make you notice awesome things about the book that you didn't appreciate the first time. The best books stand up to several readings.
If you play a roguelike, you're playing a different experience every time but it's build out of the same blocks. Once you've experienced all of those blocks - which may take a while - you start to recognize everything but you may be delighted by how those parts are presented. You begin to replay with knowledge of what's probably coming, hoping for certain outcomes, enjoying that gameplay environment. The game is still offering entertainment and the occasional surprise or delight.
Replaying games like Civilization is similar - every game has a new map and a different distribution of resources. It would be cool to see a Civ game within a specific tech period that randomized which tech advances were available! At the start of the game it decides what the tech tree will be but you always have the same "theoretical tech tree" to work off of. You'd have to hope that your research would pan out, because in that world maybe nobody ever develops Optics or The Wheel. It would be hard to do this with a game like Civilization that spans multiple tech periods because losing an early tech will frequently make later techs impossible. There could be hints as to which techs can exist based on stuff you find, or maybe research requires the same resources you will use to take advantage of that tech, so if the resource is lacking it'll not only be hard to research but once you have it you won't be able to exploit it. Point is, make it less predictable. Maybe that world has no uranium, maybe it has too much! But I don't think that's what Civilization is trying to do.
I think the games that reward repeated or deep play have other things in common. You'll probably feel like you're not just getting better at them, not just learning more about them, but really digging your toes in and experiencing them. At the end of the game you probably feel relieved, satisfied, and above all that you want to play again someday. Maybe you want to play again immediately! But with the expectation that it's not going to just be the same game.
I'll use Tetris as a counter-example. Clearly you can get better at it, and there are high tiers of player skill, and it can be enjoyable, and you can want to play again immediately. But every single game is going to be virtually the same. This type of arcade game - Bejeweled, Asteroids, Space Invaders, etc. - isn't what I'm talking about. They can be enjoyable and have value in our culture but they don't provide fresh insights and surprises. Even roguelikes run dry at some point.
As another example of games that reward deep and repeated play: fighting games. Especially ones with complex interactions between moves of different characters, blocking and countering, powering-up, affecting the arena. And only ones that offer dozens of playable characters, which makes for a very large variety of matchups. I think this is why MOBAs are so popular. Of course these aren't exactly single-player games; even if the AI is good you'll probably begin to get good at fighting it, and play against other humans is more enjoyable and enriching.
However, I don't think typical FPS games do this. Yes you have the matchups between different classes, and objectives on the field. But generally there are only a few classes or weapons and many operate similarly. You compete against other players and cooperate with the players on your team, but there are generally few counters to other players. One example would be an anti-tank mine that you place to protect an area from vehicles, but which is relatively useless against infantry which might attack you. Shooting your gun to drop someone isn't a specific counter: it's just operating your gun. It's a basic attack. As such you're less interested in what the other player is doing and more with your own status: how alone are you, are you well-supplied, are you injured. Like a MOBA, there is teamwork and picking your fights, and in addition there is the strategy of organizing a large number of teammates across a large area such as in Battlefield 1942.
I think a richer environment, lots of things to play with, lots of randomized objectives, and plenty of surprises are important for this type of game. But like a good book, if the story reveals new things every time you play it, that helps too.