SonDa5 - Great Result!
I would not expect that a LM would have that much of a delta over ICD although I would expect a degree or two better under favourable LM application. I believe that removing the IHS besides removing a layer of resistance provides a exceptionally flat surface for improved contact as noted in your Contact test in post #150.
Screw systems are pretty poor mounting systems in general on our fixed test dies we stopped using them as it was awkward to say the least to get a good mount having to alternate tightening and loosing the screws by "feel" and temp observation to jiggle into the best or most even mount for best performance, an eighth of a screw turn could cant the sink in one direction or other and significantly impact temps a degree or two. To get consistency we switched to a centre point spring mount for consistent pressure and contact.
In any event I think this is what happened in you post #159 result Your sink was cocked to one side and LM being the way it is could not fill the gap hence the higher temps. If you watch the Liquid pro application video note the amount applied, the rice sized amount used is enough to cover the area but to the Intel spec of Flat and // to .002 is only approximately 1/3 or 1/4 enough to fill to the extreme end of the spec any voids and as noted in the contact images posted here most lean to much less than ideal.
Where LM's truly shine are in situations of high contact but low pressure. All thermal compounds have have there own pressure/resistance curve similar to the one below and will vary with particle size and the amount of liquid mixed in so that say a highly liquid stock compound may reach it bond line thickness (BLT) or average particle size at 40 PSI or a more viscous compound as ICD it might be 60 PSI. On on a comparison at 40PSI two compounds would appear the same but at but at a 50% increase in pressure to 60 PSI since the stock is already at it's BLT you see little change whereas the alternate compound will drop several degrees as it closer to it's optimum BLT.
Now Look at the chart below, your LM having no particles in the mix probably hits it's best BLT at a couple of PSI perhaps 5 psi? just by being purely a liquid vs a particle porous media. Now draw a straight horizontal line all the way to 70 PSI it gets no better across the range where particle loaded compounds will continue to improve till they resolve down to their average particle size or ultimate BLT.
I would suggest that to improve ICD performance would be to tighten down the sink but I think this puts the die at some risk with any miscalculation on load balance and while I believe the final result would be close if pressure was increased within a couple of degrees in the final analysis you would be still ahead with the LM as in this situation Low pressure/High contact plays to LM's strengths.
Good test and a nice job on the whole project, 26.25C reduction is nothing to snarf at.