Unfortunately ETH has to scaffold up every token given no native token service so the analogy breaks (my opinion).
For example, if I deployed the project on ETH I would use an ERC-1155 for all the NFT collections and the Fungible Token. As a result it could be 1 approval for the user to sign to give the contract permission to interact with all tokens. Plus there would be no association step. So 2 messages (approval / interact) to sign.
By virtue of HTS - which I appreciate so very much vs standing up basic architecture every time - each minted NFT token is separate. As such if I then need to interact with my staking contract I have to ask the user to sign to grant approval of all serials of each NFT collection on top of the association step.
Net/net: it's a choice between a high friction and ultimately poor user experience, or restructuring all collections onto a single jumbo token to mimic (poorly) ERC-1155 which is a rubbish experience for other reasons.
Said differently, just because it looks like ETH in the basic case does not always make it right. I am keen to keep the benefits of HTS not force ourselves to the dark ages of rolling ERC20/721/1155 just to be more ETH like.