>No it is not, unless "defined-undefined behaviour" has now become part of Interfaces.
Deterministic observed behaviour, like it or not, will always be considered a part of the ABI.
This is why the kernel goes out of its way to preserve observe but undocumented behaviour, and one of the reasons Windows is wildly successful despite its numerous design flaws is that Microsoft agrees.
If a change breaks existing software, then it's a regression. Hand-wringing, finger-pointing, and bitter recriminations about 'proprietary crapware' are all irrelevant. Something worked. Now it doesn't.
From the comments on this it sounds like symbol versioning could be used to avoid this problem altogether, while still getting the benefit for newly built applications. Developers don't want to do this because they feel that it will benefit only proprietary software. Of course the only people harmed by this attitude are end users.
This is just yet another case where open source software chooses politics over technical excellence, which is sad but entirely unsurprising.
 Disregarding the idea that one might want to use some open source software with a similar bug that hasn't yet been fixed - most developers seem to always want to run the latest bleeding edge version of everything, and don't understand that the rest of the world isn't like that and expects existing software not to break unexpectedly.