Visual consistency across versions

[StewartCanva]

We've noticed a slight shift in text rendering between versions of cosmic-text specifically related to #384 . This seems to be an expected change.

For our use case, shifts in text positioning may be noticeable to our users since they sometimes visually line things up with text elements.

I'm trying to understand the project's approach to rendering fidelity between versions:

Is maintaining consistent text rendering between versions a goal of cosmic-text?
Or is some variation in rendering expected as the library evolves? Is there a tolerance for minor changes or variations that are considered acceptable?

How are changes that affect visual output typically handled?
Are there established approaches for versioning or communicating these changes?

Not looking to push in any particular direction, just trying to understand the project's philosophy so we can plan accordingly.

[jackpot51]

#384 was done without any config for it because it appears to be the correct behavior when compared to other systems. In the future, I would expect anything that changes rendering to be opt-in. We could potentially make rounding x advance be opt-out.