to be clear, i am not advocating for splitting the repositories further - it is already far too balkanized and spinning off a separate swift-bytebuffer repo is neither necessary nor beneficial.
i think we should reexamine the assumption here:
in my view, SPM cloning all the artifacts from swift-nio is not a major issue right now.
sure, in general, the way SPM fetches dependencies has some major performance problems. but these problems are not specific to swift-nio, and it makes no sense to micro-optimize away a swift-nio fetch in 5.9, when every project that uses macros must clone swift-syntax in order to compile.
if swift-syntax didn’t exist, sure it might make sense to evaluate tradeoffs between cloning swift-nio and cloning a smaller, separate repo. but swift-syntax does exist, it isn’t going away any time soon, and i think it is okay if the fly that is swift-nio piggybacks on the elephant that is swift-syntax.