The allele frequency definition you just rejected answers exactly that question. That definition is not intended to describe the whole of evolution; it’s intended to mark the boundaries of the subject. If it involves a change in allele frequency, we consider it evolution. If not, no. For a description of what it is that evolutionary biology studies, “descent with modification” is much better, but for demarcation, the frequency definition is good.
Why does it matter, anyway? We don’t have a clear definition of what a “particle” is – and certainly not one that’s been stable across time – and the boundaries of the field are pretty fuzzy, yet particle physicists manage to study particles anyway.