You are right!
You may have your reasons but why don't you just change the rootkey at the sample or at the instrument level? The result is exactly the same at the end: at will be resampled anyway to fit the sample rate of the audio backend. The method I suggest is to ALWAYS specify at the sample level the exact tune of the sample, see this page:
www.polyphone-soundfonts.com/documentati...-to-prepare-a-sample
Then, if you need to transpose at the instrument level, use the parameter "overriding root key" and / or the fine tuning. This this a more "natural" way to transpose and as I said, the resampling will be the same compared to a weird sample rate at the sample level.