Good point, never had a problem to place the configuration files where they belong to, be it in the project root or not, until I figured it out. That's what software configuration management is for, and SCM is obviously in place (given this is a reported issue on Github).
And if you allow me to build on the "Finally" note in your comment with a personal note:
Some _human readable_ content belongs into the root as well, e.g. a read-me, information under which conditions the code can be used (copying info) and similar, the important stuff. Otherwise it would mean it only needs to be machine read-able (which can be fine, too, especially for pure configuration management, still though a repository will most likely checked out by a human from time to time).
Uncle Bob raises this question from time to time: When you checkout a repository and look into the tree, what do you see?
The root is always an opportunity to show the projects' birds-view.