Talk to yourself. You should rather try to understand what you're posting. That'd be time well spent.
I fully understand what I am posting.
What will you eat if I prove this is not happening?
Absolutely nothing. I owe you nothing. I've educated you. It is up to you to research it further if you really want to know how these things work.
If you go back and re-read what I have been saying, you'll understand that there are multiple ways to perform pitch shifting, and not all of them would require 23ms to represent an E0. However, that doesn't change the absolute fact that an E0 is 23ms long - 24.2ms to be exact - for a single cycle.
If you're pitching shifting on a pitch detection basis - whether you use autocorrelation, zero-crossing analysis, FFT peak-tracking - then you absolutely need to have at least one full cycle of the waveform to be able to analyse the pitch. This
does not mean that latency would change from pitch to pitch. That simply isn't how it works and belies a gross misunderstanding and ignorance on your part.
If you are using a time-domain granular approach, phase-vocoder shifting, or resampling approaches, then
you're avoiding pitch detection altogether. Which means you can reduce the latency. This is what the Whammy does. It is also probably what Helix does.
As I said in a previous post - there are many ways to perform this operation. But there are always trade offs. Latency is always a factor, but some approaches are worse than others. The lower latency approaches have issues with phase reconstruction and sound quality.