I get the same math... Seems fucky but... This is assuming the sum of centripetal acceleration and gravity at the peak of the loop is zero. It may be physically possible for a cat to learn to manage a loop with such velocity but I imagine a cat wouldn't be able to maintain a stride through a zero-g portion of the loop the first time it tried it.

So, instead let's throw an assumption that the cat must maintain at minimum sum of -1g at the maxima of the loop. That may be badly phrased, assuming the cat must have at minimum a net force of at least one g between it's paws and the surface of the loop it was currently using to accelerate...

3.5 meters = 11.5 feet

Radius, so still a freaking 7 meter diameter loop feels incredible...