When you playtested the cards, how much time did you spend testing them in 2 person games vs. 3-4 person games? Did you spend a significant amount of time testing cards for 5+ player games?
I have done more playtesting with three to five than with two.
Typically my goal for a game is to support three to five, and I focus on four because, that will be close. Four will tend to be in the middle of three and five, I mean this seems obvious but probably we could find counterexamples. Anyway I focus on four. And I go for three to five because that's how many people I'll have; I don't generally have just one other person there. Now it's great to support two if you can and I find out once it's working if it does support two. Other games I may aim for two to four or just two but normally I am aiming higher because I want to deal with the actual groups I am playing with. If six is no problem then I include it, but at the high end the game tends to get slower (even with simultaneous decisions) and you start having to include too many components.
For Dominion it was immediately clear it would work with two and so overall the game aims to support two to four, which means balancing it for three (though maybe a card will cause issues at one end and I will have to deal with that). But then there I am at a game night, I am setting up the game and people want to play. If the game handles them I let them play, I don't say, "no I really want to test three-player in particular." So I have played plenty of five-player Dominion even though I prefer three and then four and then two. I decided six was too many and wouldn't bring the cards for six, to make sure I didn't get suckered into it.
I have played a lot of two-player Dominion anyway, especially online. Online we wouldn't play with five; you aren't getting as much social aspect as irl and the slower play starts to bother you more. And it's easy enough to start two separate games and then see if someone leaves or shows up.