It's another variation on the classic, "6 years experience with X required" where X is something that has only been publicly available for a year or less. Everybody wants candidates who can hit the ground running even when that's not practically possible.
This is just one example of the overall problem. In my view, hiring managers often are just going off a list of requirements and do not really know too much about the technology. They tend to reject anyone who does not use the exact terms they are looking for, even if they have a sufficient amount of experience. I'm thinking about people who worked in optimization for a long time before applying to machine learning jobs. Both fields have basically equivalent work, just with different techniques, and optimization is a lot older and more mature of a discipline. If hiring managers do not know that, they will skip over qualified candidates.
