If there is an obscure setting somewhere which makes something work well for you while you become a minority of users in the process, you would not use it?
>I never want to be outside of the primary use case.
You've extrapolated too far from their statement. Most people have probably been in this kind of situation and most of us have probably been burned by it at some point: Use a product for an edge case of its intended function and you risk losing that functionality at some point.
Vendors will generally cater for the masses. Where possible, avoid building your usage model around niche features. Assume that free products/features are a loss leader and will disappear at some point. Have a contingency plan if the at-risk feature is particularly important to you.
Depending on what you mean by "in some future update" and "pretty much guaranteed" (given an infinite timespan everything will disappear) I don't think that's true. I've kept my Firefox user.js (my manual about:config changes) under git over the past 3 years[0], and of the 44 options that I customised, 36 are still present and (seem[1] to be) active.
6 of the about:config user_prefs customised add-ons, so they no longer work due to the shift to Webextensions (but I can still make 5 of the customisations via another interface).
1 customised the GCLI, which was removed, and 1 customised Panorama, which was also removed. (However, most of what GCLI did, can be done some other way, and there are a couple of Webextensions faithfully emulating Panorama.)
[0] The file is older, but I added it to git only three years ago. Hence, many of the about:config changes have been "alive" for longer, but I have no record of those that were removed earlier than 3 years ago.
[1] Cursorily glancing through my comments above each entry to make sure that it still does what it was supposed to.