floating point errors when adding parts of entropy

[micsthepick]

because we are doing up to 255 sequential additions, and not in any particular sequence, files with the same real entropy will have slightly different entropies displayed on the program's output. I suppose to fix this issue (if you see it as a proper issue) you could sort the counts by size with an appropriate integer sorting algorithm, or maybe there's a fp-precise compiler option missing?

[micsthepick]

and I'm happy if you don't care about this issue, it just bugs me seeing stuff that mathematically should be equal but has a few wrong digits :)