the scriptfile parser already returns a pointer to a vector of string pairs - we could just use that structure as-is when loading string keys, right? can't really forsee any reason for having translation keys loaded from multiple files.
..oh, right, we'll need a script parser wrapper function anyway for good practice across the project as well as clearing out the {} brace lines from the loaded script. still, halfway up and running in short order. next is letting the rest of the project poke the keys manager. maybe string hashing as well at this point with how many strings get passed around? though we can definitely just take the perf hit and not bother. not like there's much overhead here.