If a definition of genocide is sensitive to where we mark the start and end of the genocide, then it isn't a very good definition of genocide. We can do the same thing with area: suppose some ethnic group was being genocided in a particular region, but overall population growth of that group was positive. Does that make it no longer a genocide? Clearly not.
I don't quite see where the miscommunication is. Serious claims about genocide normally come with (at least rough) temporal and geographic scopes. If we use too broad a scope, like "the Holocaust occurred from 1933-2025", then the claim becomes false. Right?
I think most people claiming a genocide is occurring are using a broad scope, like the conflict in Gaza since Oct 7 (if not something even broader), so it seems appropriate to look at the population change within that time period.
OTOH noone is claiming a genocide of Jews occurred from 1933-2025, so it wouldn't make sense to look at population change for that entire period.
The definition of the term genocide doesn't mention net population decrease - or in fact any population decrease at all. I think I've demonstrated why it can't possibly include such a requirement. The way I did so involves, exactly as you imply, assuming that the holocaust was a genocide. Then I show how such a requirement would contradict that assumption.