That's u87, measured in Neumanns QC(smaller) chamber. And yes it seems wrong even though u87 should have some hf attenuation, but not that much.
That curve will heavily depend on their calibration they use there, and even though it leaked through "How its made", it's not meant for general public. It could be relative, not absolute frequency response.
There might be another thing at play. That would be a typical response for F8. It could be they tested it in F8 mode, to be able to measure both front and back. However we don't know, it isn't obvious in the video.