I think the questions we need to ask in order to answer yours are "What do other languages/ecosystems gain from having their own centralized package repository?" and "Does this apply to C/C++?"
I feel the answer to the first question has something to do with the initial promotion of a new language: the early adopters want to make it as easy as possible for newcomers to enter the ecosystem, acquire useful, tested code and contribute back their own. For obvious reasons, the "usage graph" always has a single root - the creator(s) of the language. There's usually one reference implementation (at least initially) and therefore any code you might want to share has to conform to it.
This makes it easy to create packages that just download and compile. Certainly, had C or C++ been introduced in 2013, their communities could have followed a similar evolutionary path, but they hadn't and there's no one single prevailing toolchain to apply a package manager to. This makes the implementation of such a program too troublesome to be worth the hassle. (should you make users choose between libfoo-gcc and libfoo-vs? Do you leave it up to the packager to resolve? Or the build process? If so, how is a package any different than a straight-up tarball?)
So to sum up my answer to the first question, I think the pattern of creating package managers serves mostly to drive adoption.
With that in mind, I think it's fairly easy to see why no single system has risen to fulfill this need - because the need doesn't exist for C and C++ programmers. What does constitute a problem for the C and C++ community (or any programmer community, really) is the need originally implied: to distribute, keep up to date and contribute back code. This has been solved many times by different people with varying degrees of success, and indeed one system is gaining significant market share: git (and some other systems before that).
Basically when the problems differ, the solutions look different too, but IMHO the difference between typing
gem install and
git clone is moot.