Maybe not. I do want to make a branch for really testing this. We can get to that 90% (as I mention below).
Some assumptions from extended may not hold true now, also considering the addition of the zoom feature in the meantime.
With cross platform software, you sometimes give up making things look & work*exactly* the same for the ability to run the software on different systems. There are always idiosyncrasies you have to deal with and, since we're relying on Tk to do the canvas drawing which in turns relies on the underlying platform APIs, we don't have total control of all aspects of this. If we were rendering the GUI with something like openGL then maybe, but even that has differences to work around between systems. But working with something like openGL negates the development savings of using Tk to being with. It's a balance between "doing everything yourself" and "doing things with others."
Basically, I think we can get pretty *close* to things looking the same and, arguably, we can definitely be close with using the same font on all systems. We're pretty close now. But we don't have total control in how the fonts are rendered. Windows and macOS deal with sub pixel rendering and antialiasing in different ways, for example. What I'm saying is that spending 100% more time on that last 10% to make things "pixel perfect between all platforms" is perhaps not worth the effort and may not be completely possible without tons of fragile hacks. We can get to that 90% relatively easily but is the 10% more worth it? Especially for an open source project relying on volunteers?
This is where my devil's advocate argument comes in: perhaps fixing the 10% by padding out your help patches is an easier solution?