This probably isn't really correct. Their blog says that 94% of the traffic they were getting over Tor was malicious, and since they probably automatically put IPs where a lot of malicious traffic comes from on the CAPTCHA list, my guess is all the Tor nodes got flagged automagically.

What they went out of their way to do, was explicitly make it less painful for legitimate users to use Tor, despite the amount of malicious content they get from Tor. I'd argue for most companies, if 94% of the traffic from somewhere is malicious, the answer is "block it and be done with it", but clearly, Cloudflare actually values Tor and what it stands for enough to come up with a workaround.