Fix min_max computation when resolving flexible lengths to follow the spec

[TimJentzsch]

What problem does this solve or what need does it fill?

The min_max computation when resolving flexible lengths seems to not follow the spec:
https://github.com/DioxusLabs/sprawl/blob/11701ee97d68b22a231cac081aab4308af136706/src/flexbox.rs#L669-L675

What solution would you like?

The computation should follow the spec. Otherwise, the computation should probably be refactored.

What alternative(s) have you considered?

If it works, we could keep it as-is. Probably a bad idea though.

Additional context

Originally discussed in #39.

[Weibye]

If it works, we could keep it as-is. Probably a bad idea though.

I agree that this should be fixed. Keeping these 'ghosts' in the code makes it harder to maintain and understand where things go wrong

[nicoburns]

I believe this one was also fixed as part of #291