From some of the replies further down that thread, it appears this is not as big a problem as it first seems.

First of all, the dynamic linker is slow and not optimized for program startup times. [ https://lore.kernel.org/lkml/4b77489c05f6459ead7233c3fb69f61... ]

And second, compiling clang with Profile Guided Optimization makes a big difference. [ https://lore.kernel.org/lkml/YItU3YrFi8REwkRA@archlinux-ax16... ]