TIM is supposed to form a super thin layer, just enough to fill any voids in the interface between HS and CPU. Most also harden at least somewhat over repeated heat/cool cycles; it doesn't stay in liquid form.
We've reviewed a dozen or more TR heatsinks over the years, and I don't recall ever seeing less than excellent performance from the vast majority of them, nor of their performance changing over any period of time. I'm not positive that TR has employed this convex base design all that time, but certainly for some years.
If you found an improvement with lapping, well that's good, but I wonder what other factors were in play? There are a few, potentially --
-- less than ideal tightening of HS to CPU (for whatever reason)
-- too much/little TIM
Finally, sure, you could have had a sample with too convex a base -- or a CPU with a really flat or even convex heatspreader -- anything is possible.
I experimented with TIM and I experimented with different pressures and mounting directions, nothing made a real difference until I lapped it.
Yes I have considered that the CPU heatspreader might be at the root of the problem but that doesn't really change anything as far as the solution is concerned,
as I don't think the store would let me return my CPU because of a convex base
In all honesty its such a simple fix and I think if you are willing to mount gigantic heatsinks to your motherboard, taking an hour to lap it (very roughly no need to go mirror polish it for 2 days) really is an insignificant investment of time and risk and therefore a nobrainer.