Wow. Lots of things are happening here. I have only the most limited suggestions. To start with, I'm afraid you need a lager sample. Its just fine to do something in duplicate for added accuracy, but when your two samples a wide apart, then all bets are off. Consider doing maybe 10 samples at your worst result, 35°C, to check your precision with all your manipulations.
Wine is a complex sample, with multiple sources of acid. Consider a sort of dry run: prepare a 10% ethanol in water sample, and see if that sort of sample behaves the way you expect at various temperatures and time points.
I have no idea where you're going with some of your missive. The autoionization of water always happens, and accounts for water's slight conductivity. It will never effect a titration or a pH meter. Or at least, is always the case in every aqueous sample, so not something we have to worry about.