When new features are added, a lot of considerations have to be made about whether the way that some feature is being presented is in fact the best way to do so. I can't speak for others, but at least it is a huge consideration for me. It is in fact, what I would call the biggest barrier to actually implementing something I know exactly how to do. There are several features, I know exactly how to do, but the potential awkwardness and/or unavoidable ambiguity or unintuitive design of actually using the feature effectively prevents me from adding the feature at all. Even of features that do make it in, this aspect undoubtedly is the largest aspect that slows down it's development and causes me the greatest consternation, sometimes even to the annoyance of the person who requested it.
Why? Because once a feature's design/API is chosen, we are locked into it forever.
There are (in general) no take-backsies. Get it right the first time, or every single modder for the rest of history of FSO is going to have to deal with your poor decisions. When I mod, I constantly chafe at poor design decisions made with features. It's not always the original developer's fault, but these things really add up. Inconsistent naming schemes, bad naming in general, unintuitive behavior, similar mechanics being duplicated in different areas with slightly different designs for maximum confusion, features being in the wrong place, a feature claiming to be comprehensive for some set of situations but only covers some of them, a feature doing much more than it claims to do, the list goes on. And while experienced modders can eventually learn all of this as second-nature, it's this kind of stuff than turns away newer modders and makes FSO less appealing as a platform in general.
All of this is to say that when presented with an opportunity to change a feature for the better, where it is actually possible to tally up and notify all of the feature's users about the change then yes! I will try to make that change happen. Backwards incompatible changes are not an inherent evil, it's simply an extremely rare situation given the requirements for it to be reasonable. There is no "trend away from mod backwards compatibility being a base consideration", but you're probably right that it should not have been after the stable, which is something I definitely think we should be more ironclad about.