Game 1
9. Qxd8 Kxd8 10. 0-0-0 would have been pretty brutal.
Game 2
3. ... Be6 is probably a bit of a panic. I think your pawns were covered and Be6 blocks your e-pawn. Cramping your King comes back to bite you. In that position castling was probably more important than anything else. Even losing a pawn to castle would have been better.
Game 3
Instead of 5. e4 try Bb5. You might be able to win the middle by pinning the Knight.
Instead of 10. Bg3, maybe take his strong bishop and castle. Not sure if that is analytically better, but it seems safer.
12. hxg3, take with the f-pawn and castle. Gives you a safe king and an semi-open file for your rook
Towards the end you probably should be thinking pawn end game at that point. You have doubled pawns so you have to be very precise. Especially near the end his passed pawn is very dangerous. Once he gets that passed pawn, your main goal should be to stop that. You lost a move that could have been used to take.
Overall you are doing good. Keep going!
I've also been advised to play longer games, but the issue is not that I need more time to see things, it's that I just don't see things. Besides the fact that it takes me about 5 seconds every move to look at the board and figure out where the pieces are, it's not that I'm wasting time calculating - I'm sitting there going "I don't know what I'd even consider doing in this position", in all phases of the game.
Hell, it takes me a few minutes to understand why the engine best move is the best move and that's with the engine *telling* me I missed a tactic.
It's just really really brutal, and I want to figure it out instead of giving up.