What strategy do you use to maintain different settings of Interoperability Productions across DEV, TEST, and PROD environments?

I know the next ones:

1. Place all different settings in environment variables. You have a different .env file for each environment, and you must add some code to Production for reading and setting these values. It's good for deploying into containers, but challenging for management when we have a large production. I mean, we have many settings that can vary depending on the environment: active flag, pool size, timeouts, and so on. Not only endpoints.

2. My own case. A deployment script that saves Production settings before pulling from git the current state of Production, and restores the start settings after. In this way, each environment has those settings that were set in the Management Portal last time. The Production class is controlled by git (it keeps the default settings for added elements). Needs additional code for deployment as minus. And the content of Productions should be the same in all environments (like in the first solution, however).

3. Default Settings - official way to solve this problem, as I understand it. Not tried it, can say nothing... Is somebody using it? Does it work, useful?

4. Different production classes for each environment, I mean: TestProd.cls, DevProd.cls, etc. Not tried it yet, but it looks easy, no need for additional code, can be different content in Productions. Its obvious disadvantage: you must configure each item twice (or more times) for each Production.

What do you use? I skipped something? I would like to invite you to take part in the discussion.