Configuration variables make the code ever-so-slightly longer, and introduces additional failure points (what if the config file is missing or corrupted?) and considerations for cases that will never arise with hard-coded values (should we check that the limit is always positive? what if it is 10000 months, should we do something?).
Yes, in many cases they are an improvement, but not always. Every "best practice" must be judged individually with the question "does it apply here?".
Years ago I was working with some very small piece of a web-crawling framework, and I had to do something special for www.naver.com, Korea's biggest portal. Because the string "www.naver.com" appeared multiple times in the same function, we put that into a string constant at the beginning: something like
static const string kNaverUrl = "www.naver.com";
Because having "magic constant" sprinkled in the code is bad!
...it was only much later that I realized the stupidity. As if Naver, Korea's #1 portal, will change its domain name at a whim! Even in the unlikely case that happens (say, they get acquired), so many other things will also change that my code will be surely unusable after that.
kNaverUrl was an unnecessary detour that only served to make the code longer and (slightly) less obvious, guarding against a theoretical possibility that will never come true.
Code Review has this nasty tendency of accumulating "best practices" beyond breaking point, because nobody objects to best practices. Once in a while, these practices turn out not to be that best.
Just my two cents.