It's not that simple. The benchmark speed of the card is just moving data in a specified manner. There is no processing involved, no checking or validation, thermal throttling, etc. Cards don't work that way when you put them in the camera and want to add processing, data validation, and a check routine. At one point ProGrade went so far as to differentiate the Peak Write Speed from the Sustained Speed. And the camera has no idea what kind of card you are using and what the limits are because Nikon does not test every card. Even more important, manufacturers are free to change components of the card over time. So they need a lowest common denominator to deliver a card and speeds that are likely to work.
The actual write speed probably does not matter until you fill the buffer. Once the buffer fills, your shooting speed is limited. Until then, it's the maximum frame rate. Once the buffer fills, I found wide variation in frame rate with as slow as 2.4 fps (154 MB/s) and as fast as 5.5 fps (352 MB/s) depending on which card I used. With a fast card in the D850, I could shoot 5.5 fps with 14 bit lossless compressed RAW files indefinitely until I hit the camera limit.
You can test this yourself for any card. Just format a fresh card in the camera and then shoot a 35 second burst. Make sure your settings are appropriate to shoot at the maximum frame rate for the first few seconds and take out any SD card. The EXIF data shows when each image was taken in full seconds. Count the number of frames in each full second and write them down. You should see the camera operating at peak speed until the buffer fills, then at write speed thereafter with a little variation for some cards.
Nikon normally does not update the users manual - certainly not fully - after initial release. CFEXpress cards did not exist when the D850 was released.