I am not sure how many here have sat around with a coach and other racers and discussed FTP and testing protocols. To say the least, everyone has an opinion and we all agreed that we can generate different FTP results depending on the course and equipment. My FTP is the highest climbing a constant grade on the hoods. Flat to rolling in the drops generates a different FTP as well as on the time trial bike. The trainer is another result and the track different again.
Power meters have strength and weaknesses as well as head units. IMO, most power meters and 1 second sampling work well for constant climbing power. In a crit with lots of acceleration and fast high torque situations, 1 second sampling seems too slow and transient response of most PMs except for maybe the SRM scientific poor.
My FTP can vary 10% across different technology modalities. However, if I am told to do an FTP test, being human, I am going to choose a nice steady 4 to 5% grade on my road bike riding on the hoods where I can shoot the lights out of the number. Sadly, that number should be highly discounted to convert to an hour equivalent flat to rolling on my TT bike.