Thank you all for the replies.
The first thing I tried was tightening the screws of the cross bar mount. That made a big difference as I was able to get from 47 - 49 C down to 36 - 38 C. That was much better, but I still wanted to see what the paste patterns looked like. So I removed the heatsink and took some photos as suggested. I then cleaned up and tried reseating.
Here are 6 photos in an imgur album:http://imgur.com/a/FGtWD
If you see a lot of paste and fractal-like patterns, that's a give-away that the cooler is poorly seated.
The 1st photo shows the paste pattern on the processor. Looks like excess paste got squeezed out and even onto surrounding components (probably from the heatsink sliding around while I was working with it). Does this pic show what you mean by the fractal-like patterns, meaning the cooler was poorly seated?
The 2nd photo shows the bottom of the Kotestu heatsink.
If the paste had separated out, you'd want to remix it before applying. That said, I haven't seen a paste separate out before.
I remixed the remainder of the Kotetsu thermal paste in the plastic sleeve bag it came in. I cleaned up the processor and heatsink. And the 3rd photo shows the initial blob I put on for the reseat attempt. Does that blob look like a reasonable amount?
The initial blob was not centered nor symmetrical, so I tried manually spread it out, ending up adding more paste too. The 4th photo shows the result. Too much?
How do you tell how much tightness is enough?
When you're not able to tighten the two screws anymore: I usually tight them together, two-three turns one screw, then two-three turns the other, until they both just stop (I mean, I'm not able to make any more turns, but just a small movement of a few degrees, if any).
The 5th and 6th photos show the left and right side of the cross bar mount, with the screws tightened. The first time I did this, I tightened until the screws stopped. But I guess I was using a loose grip. This was when I was getting 47 - 49 C. When I used a stronger grip, I could still turn the screws more, until they stopped again. Then I could get 36 - 38 C. But as a test, if I used an even stronger grip, the screws could go further still. This is sort of what I was afraid of, not knowing how much is enough. I suppose a torque wrench with a recommended reading like they do for tires would be more precise, but maybe not available for computer screws, haha.
Anyway, how do the photos look? There is a gap of air space between the cross bar mount and the metal plate things below. Does that gap match your experience? They are not supposed to be flush/touching, right?
I'm not sure if it's just my imagination, but it seems like if I screw the ends down further, the bar itself is starting to slightly curve. Normal?
After all this, I checked temps again, and I see idle is running 36 - 38 C as above. So, if it was poorly seated above, does that mean it is still poorly seated now?
You can generally expect an idling CPU to run 5-8C over ambient.
So with my current ambient at 25 C, does 36 - 38 C sound too high for idling? I read that Haswell tends to run a bit hotter than Sandy Bridge.
Check if the CPU is downclocking, sometimes the power settings dont allow it, and thus high idle temps, still i would do the re setting of the Kotetsu like Steve said, but do check that the CPU is downclocking to 800mhz when not in use.
Thanks for the tip. I'm not sure how to check this, but will look into it.
By the way, unless it's defective, it's not a matter of the mounting cross-bar: on the other hand, from your description the paste would look gone bad (I don't know why), and if I were you I'd use some third party paste when re-seating the cooler (any averagely good product will work well: Arctic Silver Ceramique 2, Arctic Cooling MX4, Gelid GC-2/GC-3, Noctua NH-T1, Prolimatech PK-1/PK-2/PK-3, Thermalright Chill Factor 3, Zalman STG-2 and so on, pick the cheapest you'll find).
I may try that next. If 36 - 38 C still sounds too high, maybe a different paste would help.