It's not hard nor does it require WhatsApp's cooperation. It's fairly easy to figure out something is an (even encrypted) video call (for example by the consistency of throughput of "wanting to upload") and certainly pretty easy to use a simple algorithm to deduce what that traffic looks like.

Hiding video traffic with other traffic is hard since it's easy to assume "long persistent video-like upload with consistent throughput over something like srtp" is video and get a few false positives and no false negatives.