OK, here's my best guess based on the data you've provided and my experience with looking at these numbers.
Try using calibration factors of
0.5 0.5 0.0 2.5
Edit: or maybe a little higher on core3
0.5 0.5 0.0 2.7
Run another quick Prime small FFTs test and show us the numbers with some idle time and some time at full load.
A Quad consists of two dual cores. Core0/core1 should have similar load temps and core2/core3 should have similar load temps. The first pair of cores is typically a couple of degrees hotter at full load.
At idle, all 4 cores should be very similar. Ignore core2 at idle since it is stuck but even it will be pretty close now.
The numbers should look better and I truly believe that they are closer to the true core temperatures but without more data it's only a semi-educated guess.
Here's how my Q6600 looks when running Prime small FFTs while looking good in Mini-Mode. No stability issues during the last 13 hours of testing of version 2.70.
http://img357.imageshack.us/img357/2603/minimodejg1.png
Even as the temps go to TjMax and beyond, core2 and core3 always seem to run cooler. Maybe Intel does this on purpose or maybe those cores always make better contact with the IHS during manufacturing so heat transfer and cooling is better.