Is it computationally more heavy to put all functions into one smart contract, even though you only call one function at a time? Or is it better to split all functions into individual smart contracts?
Throw in all the functions into one smart contract, once it's accepted by the nodes it doesn't make a difference going forward.
You can only have one smart contract per account, which means if you did multiple smart contracts it will cost you multiple accounts which costs RAM each time you do so as well.
I'm guessing you'd also achieve much easier maintainability when it's all in one contract.