I'm sure this won't surprise anyone, but the 29er I tested (KTM Aera Pro) was faster overall than my rigid Dynatech, but (a surprise to me) only by about 10% overall.
On the climbs with big rocky sections in them, the 29er was about 17% faster. I noticed on the 29er I felt like I maintained my speed and floated over the rockiest section, so clearly some time got saved in those places. These were the same climbs where I was faster on my 26x2.35 set up vs 26x1.8.
However, on smoother climbs and flat pedally sections, the 29er was faster, but only by about 3%. This difference is so small it might even just be down to the faster rolling tyres (Rocket Rons) and/or the better suspension on the 29er. Again, on these sections I found no difference between 26x2.35 and 26x1.8 either.
My conclusion is that a 29er is almost certainly faster on average than a 26er, but it is possible for this speed advantage to be vanishingly small on some types of trail. Obviously, this applies to the very specific circumstances I've tested, and to me personally. But I'm glad I know what kind of boost I can expect from spending over a grand on a new 29er...
I understand now why the bike industry doesn't bother to quantify the differences between the different wheel sizes on offer, because it depends so much on the specifics of the trail and terrain.