Well, this is complicated, and I'm not really strong enough to answer, but since nobody else has and it's an interesting question... My understanding is that there can indeed be a problem with castling early in the Italian, and that first game illustrates it: a bishop pinning the knight to the queen can be a real pain (often it makes it difficult for Black to play ...d5, which would equalise a lot of the time). You might compare it to a line like this:
Obviously, this ...h6 ...g5 stuff is rather riskier if you've already castled kingside. Sometimes it might be playable, but it exposes your king somewhat, and you always have to calculate the consequences of a sacrifice on g5, as played by Kramnik.
I don't suppose castling in the position Howell did should lose or anything (I see a few high-level games in my database where it happened and black turned out okay), but that pin can be annoying, and it's notable that in many games Black delays castling until White has played Nbd2, ruling out Bg5.
I should add that in lines where White plays d4 instead of d3 castling tends to be more urgent, because the centre is getting opened and you don't want your king there when that happens. But the lines with d3 tend to be quieter (at least for a while), so Black can afford to make other useful moves first.
Is it wrong for Black to castle before White? I figured that Black would be able to find counterplay in the center but in these games it seemed ineffective.
So if White does not castle then I figure that Black should not either. And I managed to find such a game.
Still I really don't understand what's going on or why in the other game Black's ...d5 seemed to do nothing.