Remove layout_flexbox function from prelude

[Weibye]

Objective

The prelude contained a single method that has been kept around for backwards compatibility. It is simply a wrapper around FlexboxAlgorithm::perform_layout(). The idiomatic way to use taffy now and going forward is to use the perform_layout() on the trait so I don't think we need to keep this around for much longer.

The migration path is also very simple, but it is a breaking change.

Feedback wanted

Are there any reasons why we would not want this? Or want to delay this? I know #326 might change how end users interact with taffy, and so we might want to bundle these changes together?

[alice-i-cecile]

I like these changes. @nicoburns, what are your plans for the next breaking release? I think we should probably hold off on this PR until we have a better reason for a breaking release.

[nicoburns]

I like these changes. @nicoburns, what are your plans for the next breaking release? I think we should probably hold off on this PR until we have a better reason for a breaking release.

My plan is roughly to include the following:

• Optimise flexbox layouts with min/max sizes #413 (non-breaking)
• WIP: Support overflow property #412 (probably overflow: hidden and overflow: scroll but not overflow: auto)
• Key parts of WIP: Refactor the LayoutTree trait #326 (split into separate PRs)

I'm pretty happy to accept breaking changes into main at this point (but agree that we should hold off on actually releasing them). We have the 0.3.x branch that we can use to backport non-breaking bugfixes if need be (@alice-i-cecile - did you ever make those config changes to disable "merge committing" PRs but enable pushing directly to 0.x.y branches for maintainers?)

---

Finally, I would note that I'm not sure that the current FlexboxAlgorithm-like interfaces make sense either, as I've found it awkward to make all of the layout algorithms conform to a single LayoutAlgorithm trait. So I would expect there to be further changes to the "low-level" public API prior to a 0.4 release. I think that being the case actually makes now a good time to remove this method though, as we'll be making breaking changes in this area anyway.

[nicoburns]

Rustfmt and markdown lints need to be fixed.