A lot of Data Centers do not use 10GB copper. Instead they use 10GB Fiber which still is rather expensive for consumers. So why suggest the 10GB copper option?
Because in the consumer space, fibre is a very expensive option and most people don't have the gear to connect to it so it's another expense to upgrade devices. It'll remain this way for the foreseeable future.
For the extra few cents per metre, you can roll out Cat 6 as opposed to Cat 5/5e and future-proof the installation. When (not if) 10 GigE becomes available in more consumer devices, people can just plug in and go essentially. It happened with 10 -> 100 Mbps and again at 100 -> 1000 Mbps.
If you're rolling out a copper network, Cat 6 is really a "no brainer". Unless for some reason you have a very tight budget or have rolls of Cat 5e sitting around that you can use and costs you nothing.
In practical terms. What do you consider as "longer runs" in regards to CAT5 and CAT 6?
When you start getting into 10's of metres, that's when you start seeing effects of noise, depending on the enviroment.
How do you measure the actual performance of the network. Do you use tools like iperf and whats your experience in regards to loss of performace when using CAT 5 instead of CAT 6. My results show an average of 940Mbit/s on a CAT 5 cabling (1Gbit/s) with cable runs of 20-25 meters.
Network analysers usually, but you can do real-world tests just by using your favourite file copy utility. I'm not disputing your results, but you need to remember that in a consumer environment, the installation is probably going to be less-than-ideal when compared to commercial installs.
In a residential home, you generally don't have the luxury of dedicated data conduits and cable trays, usually people want their Ethernet points near power so you're going to come across some mains power cabling.