I think you may be, severely, over estimating the quality of contributions that come in from random people to large projects hosted on Github. The pull request process tends to be highly iterative. Someone sends in the request, we send back suggestions, they make revisions - the back-and-forth continues until a final state is arrived upon. The nasty code that existed in the interim is really not of any use to anyone (and that's saying nothing of their general ability to write commit messages, which is generally lacking).

Anyone that demonstrates a repeated ability to make pull requests that require no revisions (and that have commit messages that are useful) are just given straight-up commit access (thus this particular tool does not apply to them).