Higher taxonomy
Is it generally agreed that higher taxonomy does not, in itself, impact the concept (circumscription) and therefore different classifications of a taxon are not criteria for a concept identifier change?
Yes and no, and I think this is the trouble with determining whether the circumscription was changed or not (your point 2 or 3).
In many cases, transferring a species into a new genus does implicitly change the description, because a set of previously unrecorded or misinterpreted characters has been studied and the genus description is now implicitly correct for these transferred species.
It may be that the diagnostic characters are still the same, it may be that previous option to misapply a name (previously undetected) now no longer apply.
For example, conidiogenesis in imperfect fungi is now routinely observed, and transferring a species into a genus with defined conidiogenesis in the last 50 years generally meant that the range of potential identification (correct or misapplied) was drastically changed. This occurs on all higher levels, not only but including genus.
I have therefore doubts whether the distinction between 2 and 3 should be made a priori; a system where a name change requires a new ID and where the reasoning is based on these IDs could be more flexible to adjust.
Gregor