The main issue in the blog post is about "you, interviewee, which PL to use for coding interview" and the statistics OP uses is about what PL people uses on CodeEval (which is a place people go on to challenge themselves).

So you are right about the doubt. Simply put, a lot of interviews tend to be fixed. Interviewees don't get to choose the PL to use during coding session. It is usually fixed. If you are interviewing for Java software engineer candidate, then coding in Python makes no sense. So that statistics does not reflect reality. It's merely to say that "Python and Ruby alike PL" are popular should the candidate has the right to choose a PL.